/// <summary>
/// 获取公众号微信号wechatid的信息 (实际上就是搜索公众号id，获取第一个结果)
/// wechatid: 公众号id
/// 因为wechatid唯一确定，所以第一个就是要搜索的公众号
///
/// </summary>
/// <param name="wechatid"></param>
/// <returns></returns>
public OfficialAccount GetAccountInfoById(string wechatid)
{
//"""
OfficialAccount info = this.SearchOfficialAccount(wechatid, 1)[0]; //可能为空
return(info);
// return info[0] if info else False
}
/// <summary>
/// 最近文章页 公众号信息
/// </summary>
/// <param name="text"></param>
/// <param name="url"></param>
/// <returns></returns>
public OfficialAccount _ResolveOfficialAccount(string htmlText, string url)
{
OfficialAccount officialAccount = new OfficialAccount();
officialAccount.AccountPageurl = url;
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(htmlText);
var profileInfoArea = doc.DocumentNode.SelectSingleNode("//div[@class='profile_info_area']");
officialAccount.ProfilePicture = profileInfoArea.SelectSingleNode("div[1]/span/img").GetAttributeValue("src", "");
officialAccount.Name = profileInfoArea.SelectSingleNode("div[1]/div/strong/text()").InnerText.Trim();
//name = Tools.replace_space(name);
string wechatId = profileInfoArea.SelectSingleNode("div[1]/div/p/text()").InnerText.Trim();
if (wechatId.Length > 0)
{
wechatId = wechatId.Replace("微信号: ", "");
}
else
{
wechatId = "";
}
officialAccount.WeChatId = wechatId;
officialAccount.Introduction = profileInfoArea.SelectSingleNode("ul/li[1]/div/text()").InnerText;
string authInfo = profileInfoArea.SelectSingleNode("ul/li[2]/div/text()").InnerText;
if (!string.IsNullOrEmpty(authInfo))
{
officialAccount.IsAuth = true;
}
else
{
officialAccount.IsAuth = false;
}
string qrcode = WebUtility.HtmlDecode(doc.DocumentNode.SelectSingleNode("//*[@id='js_pc_qr_code_img']").GetAttributeValue("src", ""));
if (qrcode.Length > 0)
{
qrcode = "http://mp.weixin.qq.com" + qrcode;
}
else
{
qrcode = "";
}
officialAccount.QrCode = qrcode;
return(officialAccount);
}
/// <summary>
/// 搜索公众号
/// </summary>
/// <param name="keyword">搜索关键字</param>
/// <param name="page">第几页</param>
/// <returns></returns>
public List <OfficialAccount> SearchOfficialAccount(string keyword, int page = 1)
{
List <OfficialAccount> accountList = new List <OfficialAccount>();
string text = this._SearchAccount_Html(keyword, page);
HtmlDocument pageDoc = new HtmlDocument();
pageDoc.LoadHtml(text);
HtmlNodeCollection targetArea = pageDoc.DocumentNode.SelectNodes("//ul[@class='news-list2']/li");
if (targetArea != null)
{
foreach (HtmlNode node in targetArea)
{
try
{
OfficialAccount accountInfo = new OfficialAccount();
//链接中包含了& html编码符，要用htmdecode，不是urldecode
accountInfo.AccountPageurl = WebUtility.HtmlDecode(node.SelectSingleNode("div/div[@class='img-box']/a").GetAttributeValue("href", ""));
//accountInfo.ProfilePicture = node.SelectSingleNode("div/div[1]/a/img").InnerHtml;
accountInfo.ProfilePicture = WebUtility.HtmlDecode(node.SelectSingleNode("div/div[@class='img-box']/a/img").GetAttributeValue("src", ""));
accountInfo.Name = node.SelectSingleNode("div/div[2]/p[1]").InnerText.Trim().Replace("<!--red_beg-->", "").Replace("<!--red_end-->", "");
accountInfo.WeChatId = node.SelectSingleNode("div/div[2]/p[2]/label").InnerText.Trim();
accountInfo.QrCode = WebUtility.HtmlDecode(node.SelectSingleNode("div/div[3]/span/img").GetAttributeValue("src", ""));
accountInfo.Introduction = node.SelectSingleNode("dl[1]/dd").InnerText.Trim().Replace("<!--red_beg-->", "").Replace("<!--red_end-->", "");
//早期的账号认证和后期的认证显示不一样？，对比 bitsea 和 NUAA_1952 两个账号
//现在改为包含该script的即认证了
if (node.InnerText.Contains("document.write(authname('2'))"))
{
accountInfo.IsAuth = true;
}
else
{
accountInfo.IsAuth = false;
}
accountList.Add(accountInfo);
}
catch (Exception e)
{
logger.Warn(e);
}
}
}
return(accountList);
}
/// <summary>
/// GetArticleByCategoryIndex
/// </summary>
/// <param name="categoryIndex">从0开始，首页分类，热门0, 推荐:1,段子手:2,养生堂:3,私房话:4 范围0-20 ?</param>
/// <param name="page">页，从0开始</param>
/// <remarks>get_recent_article_url_by_index_single</remarks>
/// <returns></returns>
public List <Article> GetArticleByCategoryIndex(int categoryIndex, int page)
{
string pageStr = "";// "pc_" + page;
if (page == 0)
{
pageStr = "pc_" + categoryIndex; //分类N第0页格式为xxxx/pc/pc_N/pc_N.html
}
else
{
pageStr = page.ToString();
}
//http://weixin.sogou.com/pcindex/pc/pc_4/pc_4.html //分类4第0页
//http://weixin.sogou.com/pcindex/pc/pc_2/1.html //分类2第1页
//http://weixin.sogou.com/pcindex/pc/pc_3/2.html //分类3第2页
string url = "http://weixin.sogou.com/pcindex/pc/pc_" + categoryIndex + '/' + pageStr + ".html";
WebHeaderCollection headers = new WebHeaderCollection();
headers.Add("Host", "weixin.sogou.com");
headers.Add("Referer", "http://weixin.sogou.com/");
headers.Add("Accept", "*/*");
HttpHelper NetHelper = new HttpHelper();
string text = NetHelper.Get(headers, url, "UTF-8");
HtmlDocument pageDoc = new HtmlDocument();
pageDoc.LoadHtml(text);
string targetXpath = "";
if (page == 0)
{
targetXpath = "//ul[@class='news-list']/li";
}
else
{
targetXpath = "li";
}
var targetArea = pageDoc.DocumentNode.SelectNodes(targetXpath);
List <Article> ListArticle = new List <Article>();
if (targetArea != null)
{
foreach (var li in targetArea)
{
try
{
Article article = new Article()
{
Imgs = new List <string>()
};
OfficialAccount account = new OfficialAccount();
article.Title = li.SelectSingleNode("div[2]/h3/a").InnerText;
article.Url = li.SelectSingleNode("div[1]/a").GetAttributeValue("href", "");
article.Brief = WebUtility.HtmlDecode(li.SelectSingleNode("div[2]/p[@class='txt-info']").InnerText);
string coverImg = li.SelectSingleNode("div[1]/a/img").GetAttributeValue("src", "");
if (!string.IsNullOrEmpty(coverImg))
{
article.Imgs.Add(coverImg);
}
article.Time = li.SelectSingleNode("div[2]/div/span").GetAttributeValue("t", "");
article.ArticleListUrl = li.SelectSingleNode("div[2]/div/a").GetAttributeValue("href", "");
account.AccountPageurl = li.SelectSingleNode("div[2]/div/a").GetAttributeValue("href", "");
account.Name = li.SelectSingleNode("div[2]/div/a").InnerText;
string isV = li.SelectSingleNode("div[2]/div/a").GetAttributeValue("data-isv", "");
if (isV == "1")
{
account.IsAuth = true;
}
else
{
account.IsAuth = false;
}
account.ProfilePicture = li.SelectSingleNode("div[2]/div/a").GetAttributeValue("data-headimage", "");
article.officialAccount = account;
ListArticle.Add(article);
}
catch (Exception e)
{
logger.Error(e);
}
}
}
return(ListArticle);
}
/// <summary>
/// 搜索微信文章
/// </summary>
/// <param name="name"></param>
/// <param name="page"></param>
/// <returns></returns>
public List <Article> SearchArticle(string keyword, int page = 1)
{
List <Article> articleList = new List <Article>();
string text = this._SearchArticle_Html(keyword, page);
HtmlDocument pageDoc = new HtmlDocument();
pageDoc.LoadHtml(text);
//todo
HtmlNodeCollection targetArea = pageDoc.DocumentNode.SelectNodes("//ul[@class='news-list']/li");
foreach (HtmlNode node in targetArea)
{
try
{
string url = WebUtility.HtmlDecode(node.SelectSingleNode("div[2]/h3/a").GetAttributeValue("href", ""));
string title;
List <string> imgs = new List <string>();
string brief;
string time;
OfficialAccount account = new OfficialAccount();
string spans;
if (!string.IsNullOrEmpty(url))
{
title = node.SelectSingleNode("div[2]/h3/a").InnerText;
string img = WebUtility.HtmlDecode(node.SelectSingleNode("div[1]/a/img").GetAttributeValue("src", ""));
imgs.Add(img);
brief = node.SelectSingleNode("div[2]/p").InnerHtml;
time = node.SelectSingleNode("div[2]/div/span/script/text()").InnerHtml;
if (node.SelectSingleNode("div[@class='txt-box']/div[@class='s-p']/a") != null)
{
account.IsAuth = Convert.ToBoolean(Convert.ToInt16(node.SelectSingleNode("div[@class='txt-box']/div[@class='s-p']/a").GetAttributeValue("data-isv", "")));
}
account.AccountPageurl = WebUtility.HtmlDecode(node.SelectSingleNode("div[@class='txt-box']/h3/a").GetAttributeValue("href", ""));
if (node.SelectSingleNode("div/div[2]/a") != null)
{
account.ProfilePicture = WebUtility.HtmlDecode(node.SelectSingleNode("div/div[2]/a").GetAttributeValue("data-headimage", ""));
}
if (node.SelectSingleNode("div/div[2]/a") != null)
{
account.Name = node.SelectSingleNode("div/div[2]/a").InnerText;
}
}
else
{
url = WebUtility.HtmlDecode(node.SelectSingleNode("div/h3/a").GetAttributeValue("href", ""));
title = node.SelectSingleNode("div/h3/a").InnerText;
HtmlNodeCollection spansNodeCollection = node.SelectNodes("div/div[1]/a");
foreach (HtmlNode span in spansNodeCollection)
{
string img = WebUtility.HtmlDecode(span.SelectSingleNode("span/img/@src").InnerText);
if (!string.IsNullOrEmpty(img))
{
imgs.Add(img);
}
}
brief = node.SelectSingleNode("div/p").InnerText;
time = node.SelectSingleNode("div/div[2]/span/script/text()").InnerText;
if (node.SelectSingleNode("div/div[2]/a") != null)
{
account.IsAuth = Convert.ToBoolean(node.SelectSingleNode("div/div[2]/a").GetAttributeValue("data-isv", ""));
}
account.AccountPageurl = WebUtility.HtmlDecode(node.SelectSingleNode("div/div[2]/a").GetAttributeValue("href", ""));
account.ProfilePicture = WebUtility.HtmlDecode(node.SelectSingleNode("div/div[2]/a").GetAttributeValue("data-headimage", ""));
if (node.SelectSingleNode("div/div[2]/a") != null)
{
account.Name = node.SelectSingleNode("div/div[2]/a").InnerText;
}
}
if (!string.IsNullOrEmpty(title))
{
title = title.Trim().Replace("<!--red_beg-->", "").Replace("<!--red_end-->", "").Replace("<em>", "").Replace("</em>", "");;
}
else
{
title = "";
}
if (!string.IsNullOrEmpty(brief))
{
brief = brief.Trim().Replace("<!--red_beg-->", "").Replace("<!--red_end-->", "").Replace("<em>", "").Replace("</em>", "");
}
else
{
brief = "";
}
Regex TimeRegex = new Regex(@"timeConvert\('(?<1>(\d+))'\)");
var m = TimeRegex.Match(time);
time = TimeRegex.Match(time).Groups[1].Value;
Article article = new Article();
article.Title = title;
article.Brief = brief;
article.Url = url;
article.Imgs = imgs;
article.Time = time;
article.officialAccount = account;
articleList.Add(article);
}
catch (Exception e)
{
logger.Error(e);
}
}
return(articleList);
}
