void ParserSearchPageHtml(string html, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(html);
var rows = doc.GetElementbyId("similarfiles").SelectNodes(".//div");
if (rows != null)
{
foreach (var row in rows)
{
var title = row.SelectSingleNode("span/a");
var hash = Regex.Match(title?.GetAttributeValue("href", "") ?? "", @"se/(\w{40})/", RegexOptions.IgnoreCase).GetGroupValue(1);
if (string.IsNullOrEmpty(hash))
{
continue;
}
var item = CreateResourceInfo(hash.ToUpper(), title.InnerText);
item.DownloadSize = row.SelectSingleNode("span[5]").InnerText.Trim();
item.UpdateTimeDesc = row.SelectSingleNode("span[4]").InnerText.Trim();
result.Add(item);
}
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = doc.DocumentNode.SelectSingleNode("//td[@class='cur']")?.NextSibling != null;
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var nodes = doc.DocumentNode.SelectNodes("//div[@class='search-item']");
if (nodes != null)
{
foreach (var node in nodes)
{
var item = CreateResourceInfo(
Regex.Match(node.SelectSingleNode(".//a[@class='download'][1]").GetAttributeValue("href", ""), @":([A-F\d]{40})").GetGroupValue(1),
node.SelectSingleNode(".//h3/a").InnerText
);
item.DownloadSize = node.SelectSingleNode(".//div[@class='item-bar']/span[3]/b")?.InnerText;
item.UpdateTimeDesc = node.SelectSingleNode(".//div[@class='item-bar']/span[2]/b")?.InnerText;
result.Add(item);
}
}
result.HasPrevious = doc.DocumentNode.SelectSingleNode("//div[@class='bottom-pager']/span/preceding-sibling::*") != null;
result.HasMore = doc.DocumentNode.SelectSingleNode("//div[@class='bottom-pager']/span/following-sibling::*") != null;
return(base.LoadCore(context, url, htmlContent, result));
}
void ParserSearchPageHtml(string html, ResourceSearchInfo result)
{
string rowContent;
var charPosition = 0;
while (!string.IsNullOrEmpty((rowContent = html.SearchStringTag(@"<div class=""row"">", " </div>", ref charPosition))))
{
var hash = Regex.Match(rowContent, @"hash\/([a-f\d]{40})['""]", RegexOptions.IgnoreCase).GetGroupValue(1);
if (string.IsNullOrEmpty(hash))
{
continue;
}
var item = new ResourceInfo();
item.Hash = hash.ToUpper();
item.Provider = this;
//标题
item.Title = Regex.Match(rowContent, @"title=['""](.*?)['""]", RegexOptions.IgnoreCase).GetGroupValue(1).DefaultForEmpty(".....");
item.DownloadSize = Regex.Match(rowContent, @"size"">([^<]+?)<", RegexOptions.IgnoreCase).GetGroupValue(1);
item.UpdateTime = Regex.Match(rowContent, @"date"">([^<]+?)<", RegexOptions.IgnoreCase).GetGroupValue(1).ToDateTimeNullable();
result.Add(item);
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = html.IndexOf(">Next <", StringComparison.OrdinalIgnoreCase) != -1;
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var parser = new JumonyParser();
var doc = parser.Parse(htmlContent, new Uri(ReferUrlPage));
var node = doc.Find("#archiveResult tr").Skip(1);
foreach (var row in node)
{
var title = row.FindFirstOrDefault("td.name")?.InnerText();
//var size = row.FindFirstOrDefault("td.size")?.InnerText();
var date = row.FindFirstOrDefault("td.date")?.InnerText()?.ToDateTimeNullable();
var has = Regex.Match(row.FindFirstOrDefault("td.action a:nth-child(1)").Attribute("href").AttributeValue, @"/([a-z\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
var item = CreateResourceInfo(has, title);
//item.DownloadSize = size;
item.UpdateTime = date;
result.Add(item);
}
var pager = doc.FindFirstOrDefault("div.pagination");
result.HasPrevious = pager?.FindFirstOrDefault("*:first-child")?.Name == "a";
result.HasMore = pager?.FindFirstOrDefault("*:last-child")?.Name == "a";
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = CreateHtmlDocument(htmlContent);
var rows = doc.DocumentNode.SelectNodes("//table[@class='tl']/tr");
if (rows != null)
{
foreach (var row in rows)
{
var titlelink = row.SelectSingleNode("td[3]/a");
var title = titlelink.InnerText;
var linkinfo = Regex.Match(titlelink.GetAttributeValue("href", ""), @"torrent/(\d+)/(.*?)\.html", RegexOptions.IgnoreCase);
var size = row.SelectSingleNode("td[4]").InnerText;
var res = CreateResourceInfo(null, title);
res.SiteData = new SiteInfo()
{
SiteID = linkinfo.GetGroupValue(1).ToInt64(),
PageName = linkinfo.GetGroupValue(2)
};
res.DownloadSize = size;
result.Add(res);
}
var currentPage = doc.DocumentNode.SelectSingleNode("//b[@class='pager_no_link']");
result.HasPrevious = currentPage?.PreviousSibling != null;
result.HasMore = currentPage?.NextSibling != null;
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 加载数据
/// </summary>
/// <param name="key"></param>
/// <param name="pagesize"></param>
/// <param name="pageindex"></param>
/// <returns></returns>
public override IResourceSearchInfo Load(string key, SortType sortType, int sortDirection, int pagesize, int pageindex = 1)
{
var result = new ResourceSearchInfo(this)
{
Key = key,
PageIndex = pageindex,
PageSize = null
};
if (sortType == SortType.PubDate || sortType == SortType.FileSize)
{
result.SortType = sortType;
}
var url = GetSearchUrl(key, sortType, sortDirection, pagesize, pageindex);
var html = NetworkClient.Create <string>(HttpMethod.Get, url, _referUrl).Send();
if (!html.IsValid())
{
return(null);
}
//分析
var strpos = 0;
string row;
while (!string.IsNullOrEmpty((row = html.Result.SearchStringTag("td class=\"idx\"", "</pre>", ref strpos))))
{
row = System.Web.HttpUtility.HtmlDecode(row);
var hash = Regex.Match(row, @"info_hash=([a-f\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1).ToUpper();
if (string.IsNullOrEmpty(hash))
{
continue;
}
var item = new ResourceInfo();
item.Hash = hash;
item.Provider = this;
//标题
item.Title = Regex.Match(row, "torrent_name\".*?><.*?>(.*?)</a>", RegexOptions.IgnoreCase).GetGroupValue(1).DefaultForEmpty(".....");
item.DownloadSize = Regex.Match(row, @"大小:</span>.*?attr_val"">([\d\.]+( |\s)+(\w+))", RegexOptions.IgnoreCase).GetGroupValue(1);
//</span><span class="attr_val">179</span>
item.FileCount = Regex.Match(row, @"文件数:</span>.*?attr_val"">([\d\.,\s]+)", RegexOptions.IgnoreCase).GetGroupValue(1).ToInt32();
item.UpdateTimeDesc = Regex.Match(row, @"添加时间:</span>.*?attr_val"">([\d\.]+( |\s)+(\w+))", RegexOptions.IgnoreCase).GetGroupValue(1);
result.Add(item);
}
result.HasPrevious = html.Result.IndexOf("\">← 上一个", StringComparison.OrdinalIgnoreCase) != -1;
result.HasMore = html.Result.IndexOf("\">下一个", StringComparison.OrdinalIgnoreCase) != -1;
return(result);
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent"></param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
if (context.Response.Status == HttpStatusCode.MovedPermanently)
{
//Host切换了？
var host = context.Response.Redirection.Current.Host;
if (host != Host)
{
Host = host;
result.RequestResearch = true;
}
return(false);
}
var doc = CreateHtmlDocument(context.Result);
var rows = doc.DocumentNode.SelectNodes("//table[@class='data']//tr[position()>1]");
if (rows == null)
{
return(false);
}
foreach (var row1 in rows)
{
var magnet = row1.SelectSingleNode(".//a[contains(@title, 'magnet')]")?.GetAttributeValue("href", "");
var hash = Regex.Match(magnet, @"btih:([a-z\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
var titlelink = row1.SelectSingleNode(".//a[contains(@class,'cellMainLink')]");
var title = titlelink.InnerText;
var size = row1.SelectSingleNode("td[2]").InnerText.Trim();
var files = row1.SelectSingleNode("td[3]").InnerText.Trim().ToInt32();
var age = row1.SelectSingleNode("td[4]").InnerText.Trim();
var res = CreateResourceInfo(hash, title);
res.FileCount = files;
res.DownloadSize = size;
res.UpdateTimeDesc = age;
//sitedata
//kingsman-the-secret-service-2014-hc-hdrip-xvid-ac3-etrg-por-talamasca32-srt-t10610377.html
var href = titlelink.GetAttributeValue("href", "");
var info = Regex.Match(href, @"/(.+?)-t(\d+)\.html", RegexOptions.Singleline | RegexOptions.IgnoreCase);
res.SiteData = new SiteInfo()
{
PageName = info.GetGroupValue(1),
SiteID = info.GetGroupValue(2).ToInt64()
};
result.Add(res);
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = !doc.DocumentNode.SelectSingleNode("//div[contains(@class,'pages')]/a[last()]").GetAttributeValue("href", "").IsNullOrEmpty();
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
if (context.IsRedirection && context.Redirection.Orginal.Host != context.Redirection.Current.Host)
{
Host = context.Redirection.Current.Host;
result.RequestResearch = true;
return(false);
}
//分析
var strpos = 0;
string row;
//var nameMap = Regex.Matches(html.Result, @"map\[['""]([a-z\d]{40})[""']\].*?['""](.*?)[""'];", RegexOptions.IgnoreCase)
// .Cast<Match>().Where(s => s.Success).ToDictionary(s => s.Groups[1].Value, s => s.Groups[2].Value);
while (!string.IsNullOrEmpty((row = context.Result.SearchStringTag("<div class=\"r\">", "磁力链接</span>", ref strpos))))
{
row = System.Web.HttpUtility.HtmlDecode(row);
var hash = Regex.Match(row, @"/t/([a-f\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1).ToUpper();
if (string.IsNullOrEmpty(hash))
{
continue;
}
var item = new ResourceInfo();
item.Hash = hash;
item.Provider = this;
//标题
item.Title = Regex.Match(row, "<h5.*?>(.*?)</h5>", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1);
if (string.IsNullOrEmpty(item.Title))
{
continue;
}
item.DownloadSize = Regex.Match(row, @"大小.*?>([\d\.]+\s*[^<\s]+)", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1);
item.FileCount = Regex.Match(row, @"文件数：.*?>([\d\.\,]+\s*)", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1).ToInt32();
item.UpdateTime = Regex.Match(row, @"收录时间.*?>([\d\-]+)", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1).ToDateTimeNullable();
result.Add(item);
}
result.TotalPages = Regex.Match(context.Result, @"totalPages\s*:\s*(\d+)", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1).ToInt32();
result.HasPrevious = result.PageIndex > 1;
result.HasMore = result.PageIndex < result.TotalPages.Value;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 加载数据
/// </summary>
/// <param name="key"></param>
/// <param name="pagesize"></param>
/// <param name="pageindex"></param>
/// <returns></returns>
public override IResourceSearchInfo Load(string key, SortType sortType, int sortDirection, int pagesize, int pageindex = 1)
{
var url = $"http://api.xhub.cn/api.php?op=search_list&callback=jQuery{DateTime.Now.Ticks}&key={UE(key)}&page={pageindex}&_={DateTime.Now.ToJsTicks()}";
var refer = GetSearchUrl(key, sortType, sortDirection, pagesize, pageindex);
var result = CollectionUtility.CreateAnymousDictionary("", new { day = DateTime.Now, hits = 0, size = "", title = "" });
var ctx = NetworkClient.Create(HttpMethod.Get, url, refer, result: new
{
data = result,
total = 0,
status = 0
}).Send();
if (!ctx.IsValid())
{
return(null);
}
var items = new ResourceSearchInfo(this)
{
PageIndex = pageindex,
PageSize = 30,
TotalPages = (int)Math.Ceiling(ctx.Result.total / 30.0)
};
items.HasMore = pageindex < (items.TotalPages ?? 1);
if (ctx.Result.data != null)
{
foreach (var item in ctx.Result.data)
{
var res = CreateResourceInfo(item.Key, item.Value.title);
res.UpdateTime = item.Value.day;
res.DownloadSize = item.Value.size;
items.Add(res);
}
}
if (base.LoadCore(null, url, null, items))
{
return(items);
}
return(null);
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var document = CreateHtmlDocument(htmlContent);
var tbody = document.GetElementbyId("data_list");
if (tbody == null)
{
return(false);
}
var nodes = tbody.SelectNodes("tr");
foreach (var node in nodes)
{
var tds = node.SelectNodes("td");
if (!(tds?.Count > 2))
{
continue;
}
var titleLink = tds[2].SelectSingleNode("a");
var title = titleLink.InnerText;
var href = titleLink.GetAttributeValue("href", "");
//show.php?hash=a86af7382b7f8fe4852cce9f98c29344c9d22993
var hash = Regex.Match(href, @"hash=([a-f\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
var size = tds[3].InnerText;
var time = tds[0].InnerText.Trim();
var item = CreateResourceInfo(hash, title);
item.SupportPreivewType = PreviewType.Image;
item.UpdateTimeDesc = time;
item.DownloadSize = size;
item.SiteData = new SiteInfo()
{
ProvideSiteDownload = null
};
result.Add(item);
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = document.DocumentNode.SelectSingleNode("//a[@class='nextprev']") != null;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var rows = doc.DocumentNode.SelectNodes("//ul[@class='table-list newly']/li");
if (rows != null)
{
foreach (var row in rows)
{
var link = row.SelectSingleNode("./div[@class='name']/span[1]/a");
var dlink = row.SelectSingleNode("./div[@class='name']/a[contains(@class, 'btn-download')]");
//torrent/3344360/%5BCWPBD-76%5D.Actress.Kiss.Blow.Job.HD720p.mkv/
var linkinfo = Regex.Match(link.GetAttributeValue("href", ""), "torrent/(\\d+)/([^/]+)", RegexOptions.IgnoreCase);
var title = link.InnerText;
var item = CreateResourceInfo(null, title);
var downloadlink = dlink == null ? null : $"http://www.torrents.net/{(dlink.GetAttributeValue("href", ""))}";
item.SiteData = new SiteInfo()
{
SiteID = linkinfo.GetGroupValue(1).ToInt64(),
PageName = linkinfo.GetGroupValue(2),
ProvideSiteDownload = !string.IsNullOrEmpty(downloadlink),
SiteDownloadLink = downloadlink
};
item.DownloadSize = row.SelectSingleNode("span[@class='size']")?.InnerText;
result.Add(item);
}
}
var pagebar = doc.DocumentNode.SelectSingleNode("//div[@class='pagination-bar']");
if (pagebar != null)
{
result.HasPrevious = pagebar.SelectSingleNode("a[1]/span/strike") == null;
result.HasMore = pagebar.SelectSingleNode("a[2]/span/strike") == null;
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(context.Result);
var nodes = doc.DocumentNode.SelectNodes("//table[@class='table2'][last()]//tr[position()>1]");
if (nodes == null)
{
return(false);
}
foreach (var node in nodes)
{
var a = node.SelectSingleNode("td[1]//a");
if (a == null)
{
continue;
}
var href = a.GetAttributeValue("href", "");
//game-of-thrones-season-3/6f185a9f14a6ce7a3794c5dd1ef3dd51f0dc9f71
var m = Regex.Match(href, @"/([^/]+)/([a-f\d]{40})", RegexOptions.IgnoreCase);
if (!m.Success)
{
continue;
}
var item = CreateResourceInfo(m.GetGroupValue(2), a.InnerText);
item.SiteData = new SiteInfo()
{
PageName = m.GetGroupValue(1)
};
item.UpdateTimeDesc = node.SelectSingleNode("td[2]").InnerText;
item.DownloadSize = node.SelectSingleNode("td[3]").InnerText;
result.Add(item);
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = doc.GetElementbyId("next") != null;
return(base.LoadCore(context, url, htmlContent, result));
}
void ParserSearchPageHtml(string html, ResourceSearchInfo result)
{
string bodyContent;
if (string.IsNullOrEmpty(html) || string.IsNullOrEmpty((bodyContent = html.SearchStringTag("<table class=\"list\" width=\"100%\">", "</table>"))))
{
return;
}
bodyContent = RemoveEmailProtect(bodyContent);
string rowContent;
var charPosition = 0;
while (!string.IsNullOrEmpty((rowContent = bodyContent.SearchStringTag("<tr", "</tr>", ref charPosition))))
{
var hash = Regex.Match(rowContent, @"value=['""]([a-f\d]{40})['""]", RegexOptions.IgnoreCase).GetGroupValue(1);
if (string.IsNullOrEmpty(hash))
{
continue;
}
var item = new ResourceInfo();
item.Hash = hash.ToUpper();
item.Provider = this;
//标题
var title = Regex.Match(rowContent, @"<a.*?href=['""].*?/(\d+)/([^'""]+?)\.html['""]>(.*?)</a>", RegexOptions.IgnoreCase);
item.Title = title.GetGroupValue(3).DefaultForEmpty(".....");
item.SiteData = new SiteInfo()
{
PageName = title.GetGroupValue(2),
SiteID = title.GetGroupValue(1).ToInt64()
};
item.DownloadSize = RemoveSpaceChars(Regex.Match(rowContent, @"nowrap>\s*(([\d\.]+).*?B)\s*<", RegexOptions.IgnoreCase).GetGroupValue(1) ?? "");
item.UpdateTimeDesc = Regex.Match(rowContent, "class=\"light\">([^><]+)</td>", RegexOptions.IgnoreCase).GetGroupValue(1);
result.Add(item);
}
result.HasPrevious = html.IndexOf("paging_prev", StringComparison.OrdinalIgnoreCase) != -1;
result.HasMore = html.IndexOf("paging_next", StringComparison.OrdinalIgnoreCase) != -1;
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var items = doc.DocumentNode.SelectNodes("//table[@class='table2'][2]//tr[position()>1]");
if (items != null)
{
foreach (var item in items)
{
var dlink = item.SelectSingleNode("td[1]/div/a[1]");
var tlink = item.SelectSingleNode("td[1]/div/a[2]");
var dcell = item.SelectSingleNode("td[2]");
var scell = item.SelectSingleNode("td[3]");
//hash
var hash = Regex.Match(dlink.GetAttributeValue("href", ""), @"torrent/([a-z\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
var pinfo = Regex.Match(tlink.GetAttributeValue("href", ""), @"/([^/]+?)\-(\d+)\.html", RegexOptions.IgnoreCase);
var resnode = CreateResourceInfo(hash, tlink.InnerText);
resnode.SiteData = new SiteInfo()
{
PageName = pinfo.GetGroupValue(1),
SiteID = pinfo.GetGroupValue(2).ToInt64()
};
resnode.UpdateTimeDesc = dcell.InnerText.Split('-')[0].Trim();
resnode.DownloadSize = scell.InnerText;
result.Add(resnode);
}
}
var pager = doc.DocumentNode.SelectSingleNode("//div[@class='search_stat']");
if (pager != null)
{
result.HasMore = pager.SelectSingleNode("./*[last()]")?.Name == "a";
result.HasPrevious = pager.SelectSingleNode("./*[1]")?.Name == "a";
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var nodes = doc.DocumentNode.SelectNodes("//div[@class='inner_container']//div[contains(@class,'grey_bar3')]");
if (nodes != null)
{
foreach (var node in nodes)
{
var titlelink = node.SelectSingleNode("p/a");
if (titlelink == null || titlelink.GetAttributeValue("href", "").IndexOf("torrent/", StringComparison.Ordinal) == -1)
{
continue;
}
var link = titlelink.GetAttributeValue("href", "");
//torrent/1653923478/Game+of+Thrones+S02E05+The+Ghost+Of+Harrenhal+HDTV+XviD-2HD+%5Beztv%5D
var linkreg = Regex.Match(link, @"torrent/(\d+)/(.+)", RegexOptions.IgnoreCase);
var data = new SiteInfo()
{
SiteID = linkreg.GetGroupValue(1).ToInt64(),
PageName = UD(linkreg.GetGroupValue(2))
};
var title = titlelink.InnerText;
var size = node.SelectSingleNode("span[4]").InnerText;
var item = CreateResourceInfo(null, title);
item.DownloadSize = size;
item.SiteData = data;
result.Add(item);
}
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = htmlContent.IndexOf("rel=\"nofollow\">>></a></li>") != -1;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var div = doc.GetElementbyId("serps");
if (div == null)
{
return(false);
}
var rows = div.SelectNodes("//tr[@data-key!='']");
var downloader = AppContext.Instance.DownloadServiceProviders.FirstOrDefault(s => s is IsoHuntDownloadProvider);
foreach (var row in rows)
{
var titleLink = row.SelectSingleNode("td[2]/a[1]");
var item = new ResourceInfo()
{
Title = titleLink.InnerText,
Provider = this,
PreferDownloadProvider = downloader,
UpdateTimeDesc = row.ChildNodes[4].InnerText,
DownloadSize = row.ChildNodes[5].InnerText
};
var href = titleLink.Attributes["href"].Value;
var hrefInfo = Regex.Match(href, @"torrent_details/(\d+)/([^'""]+)", RegexOptions.IgnoreCase | RegexOptions.Singleline);
item.SiteData = new SiteInfo()
{
SiteID = hrefInfo.Groups[1].Value.ToInt64(),
PageName = hrefInfo.Groups[2].Value
};
result.Add(item);
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
if (!context.IsValid())
{
return(false);
}
var doc = CreateHtmlDom(context.Request.Uri.ToString(), context.Result);
var nodes = doc.Find(@"table.search_table tr");
if (nodes != null)
{
foreach (var node in nodes.Skip(1))
{
var link = node.FindFirstOrDefault("td:last-child a");
var href = link.GetAttributeValue("href");
var infoMatch = Regex.Match(href, @"\|file\|([^\|]+)\|(\d+)\|([A-Z\d]+)\|");
var res = CreateResourceInfo(infoMatch.GetGroupValue(3), infoMatch.GetGroupValue(1), ResourceType.Ed2K);
res.DownloadSizeValue = infoMatch.GetGroupValue(2).ToInt64();
//preview
var preview = node.FindFirstOrDefault("td:nth-child(2) img");
if (preview?.GetAttributeValue("src")?.IndexOf("hexie") != -1)
{
res.SupportPreivewType = PreviewType.Image;
res.PreviewInfo = new PreviewInfo()
{
ImageUrl = $"http://thumb.donkey4u.com/{res.Hash}/thumb.jpg"
};
}
result.Add(res);
}
result.HasMore = context.Result.IndexOf("&mode=list\">下一页") != -1;
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = CreateHtmlDocument(htmlContent);
var table = doc.DocumentNode.SelectNodes("//div[@id='headerbox']/following::table")?.LastOrDefault(s => s.SelectSingleNode(".//tr[1]/th[6]") != null);
var rows = table?.SelectNodes(".//tr[position()>1]");
if (rows != null)
{
foreach (var row in rows)
{
var titlelink = row.SelectSingleNode("td[1]/a");
var title = titlelink.InnerText;
var pinfo = Regex.Match(titlelink.GetAttributeValue("href", ""), @"details/(\d+)/(.*?)\.html", RegexOptions.IgnoreCase);
if (!pinfo.Success)
{
continue;
}
var item = CreateResourceInfo(null, title);
item.SiteData = new SiteInfo()
{
SiteID = pinfo.GetGroupValue(1).ToInt64(),
PageName = pinfo.GetGroupValue(2)
};
item.UpdateTimeDesc = row.SelectSingleNode("td[2]").InnerText;
item.DownloadSize = row.SelectSingleNode("td[3]").InnerText;
result.Add(item);
}
var pagination = doc.GetElementbyId("pagination");
if (pagination != null)
{
var cp = pagination.SelectSingleNode("a[@class='selected']");
result.HasMore = cp.NextSibling != null;
result.HasPrevious = cp.PreviousSibling != null;
}
}
return(base.LoadCore(context, url, htmlContent, result));
}
void ParserSearchPageHtml(string html, ResourceSearchInfo result)
{
string bodyContent;
if (string.IsNullOrEmpty(html) || string.IsNullOrEmpty((bodyContent = html.SearchStringTag("<table class=\"table\">", "</table>"))))
{
return;
}
string rowContent;
var charPosition = 0;
while (!string.IsNullOrEmpty((rowContent = bodyContent.SearchStringTag("<tr>", "</tr>", ref charPosition))))
{
var hash = Regex.Match(rowContent, @"btih:(\w{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
if (string.IsNullOrEmpty(hash))
{
continue;
}
var item = new ResourceInfo();
item.Hash = hash.ToUpper();
item.Provider = this;
item.SiteData = new SiteInfo()
{
SiteID = Regex.Match(rowContent, @"/h/(\d+)", RegexOptions.IgnoreCase).GetGroupValue(1).ToInt64()
};
//标题
item.Title = RemoveHtmlStrings(Regex.Match(rowContent, @"<div>[\s]*<a.*?class=['""]title['""].*?>(.*?)</a>", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1).DefaultForEmpty("....."));
item.DownloadSize = RemoveSpaceChars(Regex.Match(rowContent, @"大小\s*:\s*([\d\.]+\s*[A-Z]+)", RegexOptions.IgnoreCase).GetGroupValue(1));
item.UpdateTimeDesc = Regex.Match(rowContent, @"更新时间:\s*<span.*?>(.*?)</span>", RegexOptions.IgnoreCase).GetGroupValue(1) ?? "";
item.FileCount = Regex.Match(rowContent, @"文件数\s*:\s*([\d\.,]+)", RegexOptions.IgnoreCase).GetGroupValue(1).ToInt32();
result.Add(item);
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = result.Count > 0 && html.IndexOf("<li class=\"disabled\"><a href=\"#\"> Next", StringComparison.OrdinalIgnoreCase) == -1;
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var document = CreateHtmlDocument(htmlContent);
var nodes = document.DocumentNode.SelectNodes("//table[@class='tor'][last()]//tr[position()>1]");
foreach (var node in nodes)
{
var titlelink = node.SelectSingleNode("td[1]/a");
var href = titlelink.GetAttributeValue("href", "");
var hrefm = Regex.Match(href, @"/(.*?)\-tf(\d+)\.html", RegexOptions.IgnoreCase);
if (!hrefm.Success)
{
continue;
}
var title = titlelink.InnerText;
var sid = hrefm.GetGroupValue(2).ToInt64();
var pagename = hrefm.GetGroupValue(1);
var size = node.SelectSingleNode("td[2]").InnerText;
var update = node.SelectSingleNode("td[5]").InnerText;
var resinfo = CreateResourceInfo(null, title);
resinfo.DownloadSize = size;
resinfo.UpdateTimeDesc = update;
resinfo.SiteData = new SiteInfo()
{
SiteID = sid,
PageName = pagename
};
result.Add(resinfo);
}
result.HasPrevious = result.PageIndex > 1;
result.HasMore = document.GetElementbyId("pgright") != null;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var list = doc.DocumentNode.SelectSingleNode("//table[@class='searchtable'][last()]");
if (list == null)
{
return(false);
}
var rows = list.SelectNodes(".//tr[position()>1]");
foreach (var row in rows)
{
var titleLink = row.SelectSingleNode("td[1]/a[1]");
//移除类别节点
titleLink.SelectSingleNode("span[@class='cat']")?.Remove();
var item = CreateResourceInfo(Regex.Match(titleLink.Attributes["href"].Value, "hash/([a-z\\d]{40})", RegexOptions.IgnoreCase | RegexOptions.Singleline).GetGroupValue(1), titleLink.InnerText);
item.UpdateTimeDesc = row.SelectSingleNode("td/span[contains(@class,'added')]/text()").InnerText;
item.DownloadSize = row.SelectSingleNode("td/span[contains(@class,'size')]").InnerText;
result.Add(item);
}
var pageDiv = doc.DocumentNode.SelectSingleNode("//div[@class='pagediv']/ul");
if (pageDiv != null)
{
result.HasPrevious = pageDiv.SelectSingleNode("li[1]/a") != null;
result.HasMore = pageDiv.SelectSingleNode("li[3]/a") != null;
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var rootNode = doc.GetElementbyId("torrents");
if (rootNode == null)
{
return(true);
}
var items = rootNode.SelectNodes("li");
foreach (var item in items)
{
var titlelink = item.SelectSingleNode("a[1]");
var href = Regex.Match(titlelink.Attributes["href"].Value, "/([^\\s\'\"]+?)\\-q(\\d+)\\.html", RegexOptions.IgnoreCase | RegexOptions.Singleline);
var descNode = item.SelectSingleNode(".//table//td[1]");
var size = descNode.SelectSingleNode("text()")?.InnerText ?? "";
var filecount = Regex.Match(descNode.SelectSingleNode("div").InnerText ?? "", @"(\d+)\s*files").GetGroupValue(1).ToInt32();
var res = CreateResourceInfo(null, titlelink.InnerText);
res.DownloadSize = size;
res.FileCount = filecount;
res.SiteData = new SiteInfo()
{
PageName = href.GetGroupValue(1),
SiteID = href.GetGroupValue(2).ToInt64()
};
result.Add(res);
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var nodes = doc.DocumentNode.SelectNodes("//article/table[last()]//tr[position()>1]");
if (nodes != null)
{
foreach (var node in nodes)
{
var titlelink = node.SelectSingleNode("td[1]//a");
//torrent/2733310/1pondo-011014_734-hd.html
var siteinfoReg = Regex.Match(titlelink.GetAttributeValue("href", ""), @"torrent/(\d+)/(.*?)\.html");
var item = CreateResourceInfo(null, titlelink.InnerText);
item.SiteData = new SiteInfo()
{
SiteID = siteinfoReg.GetGroupValue(1).ToInt64(),
PageName = siteinfoReg.GetGroupValue(2),
ProvideSiteDownload = true,
SiteDownloadLink = $"https://www.torlock.com/tor/{siteinfoReg.GetGroupValue(1).ToInt64()}.torrent"
};
item.UpdateTimeDesc = node.SelectSingleNode("td[2]")?.InnerText;
item.UpdateTime = item.UpdateTimeDesc.ToDateTimeNullable();
item.DownloadSize = node.SelectSingleNode("td[3]")?.InnerText;
result.Add(item);
}
}
result.HasPrevious = doc.DocumentNode.SelectSingleNode("//div[@id='pag']/a[@class='sel']/preceding-sibling::*") != null;
result.HasMore = doc.DocumentNode.SelectSingleNode("//div[@id='pag']/a[@class='sel']/following-sibling::*") != null;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
//分析
var content = context.Result.SearchStringTag("<div class=\"SimpleAcceptebleTextAds\">", "Web search results");
if (!content.IsNullOrEmpty())
{
var matches = Regex.Matches(content, @"<dl><dt><a.*?/([a-z\d]{40})"".*?>(.*?)</a>.*?<span\sclass=""a"">.*?title=""([^""]+)"".*?class=""s"">([^<>]+)<.*?</dl>", RegexOptions.Singleline | RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var item = new ResourceInfo();
item.Hash = match.GetGroupValue(1).ToUpper();
item.Provider = this;
item.Title = match.GetGroupValue(2);
item.DownloadSize = match.GetGroupValue(4);
item.UpdateTime = match.GetGroupValue(3).ToDateTimeNullable();
result.Add(item);
}
result.HasPrevious = context.Result.IndexOf("\">« Previous") != -1;
result.HasMore = context.Result.IndexOf("\">Next »") != -1;
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
if (context.IsRedirection)
{
var nHost = context.Redirection.Current.Host;
if (nHost != Host)
{
Host = nHost;
result.RequestResearch = true;
}
return(false);
}
htmlContent = DecodePage(htmlContent);
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
//节点
var nodes = doc.DocumentNode.SelectNodes(@"//div[@class='list-con']/dl");
if (nodes != null)
{
foreach (HtmlNode node in nodes)
{
var titleLink = node.SelectSingleNode("dt/a");
if (titleLink == null)
{
continue;
}
//pageid;http://btkitty.org/item/BcGBAQAgBATAlco_MY7Q_iN0t40SBXlwtA7jRqUsZ_no4YHR0qQ_.html
var pageinfo = Regex.Match(titleLink.GetAttributeValue("href", ""), @"/([^/]+?)/([^/]+?)\.html", RegexOptions.IgnoreCase);
var pageid = pageinfo.GetGroupValue(2);
_itemPageKey = pageinfo.GetGroupValue(1).DefaultForEmpty(_itemPageKey);
var title = titleLink.InnerText;
//detail spans.
var detailSpans = node.SelectNodes("dd[@class='option']/span");
//hash
var hash = Regex.Match(detailSpans[0].SelectSingleNode("a").GetAttributeValue("href", ""), @"([A-F\d]{40})").GetGroupValue(1);
//date
var date = detailSpans[1].SelectSingleNode("b").InnerText;
//size
var size = detailSpans[2].SelectSingleNode("b").InnerText;
//fc
var fc = Regex.Match(detailSpans[3].SelectSingleNode("b").InnerText, @"(\d+)").GetGroupValue(1).ToInt32();
var item = CreateResourceInfo(hash, title);
item.FileCount = fc;
item.DownloadSize = size;
item.UpdateTimeDesc = date;
item.SiteData = new SiteInfo()
{
SiteID = 0L,
PageName = pageid,
ProvideSiteDownload = true,
SiteDownloadLink = $"http://storebt.com/down/{pageid}/torrent.torrent"
};
result.Add(item);
}
}
//hasmore?
result.HasPrevious = result.PageIndex > 1;
var totalPage = Regex.Match(htmlContent, @"<span>共\s*(\d+)\s*页</span>", RegexOptions.IgnoreCase).GetGroupValue(1).ToInt32();
result.HasMore = totalPage > result.PageIndex;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
if (IsHostChanged(context))
{
result.RequestResearch = true;
return(false);
}
if (IsBotCheckNeeded(context))
{
if (RequireBotCheck())
{
result.RequestResearch = true;
}
return(false);
}
var doc = CreateHtmlDom(url, htmlContent);
var rows = doc.Find("table.lista2t tr").Skip(1);
foreach (var row in rows)
{
var titlelink = row.FindFirstOrDefault("td:nth-child(2) a");
var titleHref = titlelink.GetAttributeValue("href");
//如果是提供的直接下载，则忽略。这个是广告。
if (titleHref.IsNullOrEmpty() || titleHref.IndexOf("torrent") == -1)
{
continue;
}
//基本信息
///torrent/zxv627d
var sidstr = Regex.Match(titlelink.GetAttributeValue("href"), @"torrent/([\da-z]+)", RegexOptions.IgnoreCase).GetGroupValue(1);
var hash = Regex.Match(titlelink.GetAttributeValue("onmouseover") ?? "", @"/([a-z\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
if (hash.IsNullOrEmpty())
{
continue;
}
var title = titlelink.InnerText();
var size = row.FindFirstOrDefault("td:nth-child(4)").InnerText();
var date = row.FindFirstOrDefault("td:nth-child(3)").InnerText().Trim().ToDateTimeNullable();
var item = CreateResourceInfo(hash, title);
item.DownloadSize = size;
item.UpdateTime = date;
item.SiteData = new SiteInfo()
{
SiteID = 0L,
PageName = sidstr,
ProvideSiteDownload = true,
SiteDownloadLink = $"https://{Host}/download.php?id={sidstr}&f={title}.torrent"
};
//预览信息
var mouseout = titlelink.GetAttributeValue("onmouseover");
if (!mouseout.IsNullOrEmpty())
{
var imgsrc = Regex.Match(mouseout, @"<img.*?src=.*?//([^'""\\]+)").GetGroupValue(1);
if (!imgsrc.IsNullOrEmpty())
{
item.SupportPreivewType = PreviewType.Image;
item.PreviewInfo = new PreviewInfo()
{
ImageUrl = $"http://dyncdn.me/posters2/{hash[0]}/{hash}.jpg"
};
}
}
result.Add(item);
}
var pager = doc.FindFirstOrDefault("div#pager_links");
if (pager != null)
{
result.HasPrevious = pager.FindFirstOrDefault("*:first-child")?.Name == "a";
result.HasMore = pager.FindFirstOrDefault("*:last-child")?.Name == "a";
}
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
//节点
var nodes = doc.DocumentNode.SelectNodes(@"//div[@class='item']");
if (nodes != null)
{
foreach (HtmlNode node in nodes)
{
var titleLink = node.SelectSingleNode("div[@class='item-title']/a");
if (titleLink == null)
{
continue;
}
//pageid;http://diggbt.com/item/BcEBDQAwCAMwSx9h5HbGOP4lvO206MSCOsT1qDCMjiKbAz0Z3g8.html
var pageinfo = Regex.Match(titleLink.GetAttributeValue("href", ""), @"/([^/]+?)/([^/]+?)\.html", RegexOptions.IgnoreCase);
var pageid = pageinfo.GetGroupValue(2);
_itemPageKey = pageinfo.GetGroupValue(1).DefaultForEmpty(_itemPageKey);
var title = titleLink.InnerText;
//detail spans.
var detailSpans = node.SelectNodes("div[@class='item-detail']/span");
//hash
//var hash = Regex.Match(detailSpans[0].SelectSingleNode("a").GetAttributeValue("href", ""), @"([A-F\d]{40})").GetGroupValue(1);
var hash = Regex.Replace(Regex.Match(detailSpans[0].SelectSingleNode("script").InnerText, @":([a-f'""\+\d]+)&", RegexOptions.IgnoreCase).GetGroupValue(1), @"['""+]", "").Trim('&', ':');
//date
var date = detailSpans[1].SelectSingleNode("b").InnerText;
//size
var size = detailSpans[2].SelectSingleNode("b").InnerText;
//fc
var fc = Regex.Match(detailSpans[3].SelectSingleNode("b").InnerText, @"(\d+)").GetGroupValue(1).ToInt32();
var item = CreateResourceInfo(hash, title);
item.FileCount = fc;
item.DownloadSize = size;
item.UpdateTimeDesc = date;
item.SiteData = new SiteInfo()
{
SiteID = 99999999999L,
PageName = pageid,
ProvideSiteDownload = true,
SiteDownloadLink = $"http://storebt.com/down/{pageid}/torrent.torrent"
};
result.Add(item);
}
}
//hasmore?
result.HasPrevious = result.PageIndex > 1;
var totalPage = Regex.Match(htmlContent, @"<span>共\s*(\d+)\s*页</span>", RegexOptions.IgnoreCase).GetGroupValue(1).ToInt32();
result.HasMore = totalPage > result.PageIndex;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var doc = CreateHtmlDocument(context.Result);
var seedTable = doc.DocumentNode.SelectSingleNode("//table[contains(@class,'seedtable')]");
if (seedTable == null)
{
return(false);
}
var nodes = seedTable.SelectNodes(".//tr[position()>2 and position()<last()]");
if (nodes == null)
{
return(true);
}
foreach (var node in nodes)
{
var a = node.SelectSingleNode("td[2]/a");
var title = a.InnerText;
var sid = Regex.Match(a.GetAttributeValue("href", ""), @"(?<=/)\d+(?=\.html)", RegexOptions.IgnoreCase).Value.ToInt32();
var sizestr = node.SelectSingleNode("td[4]").InnerText.Trim();
var downloads = node.SelectNodes("td[3]/a");
var hasmulti = downloads.Count > 1;
//parent?
ResourceInfo parent = null;
List <ResourceInfo> subRes = null;
if (hasmulti)
{
parent = new ResourceInfo()
{
Title = title,
HasSubResources = true,
SiteData = new SiteInfo()
{
SiteID = sid
},
DownloadSize = sizestr,
Provider = this,
ResourceType = ResourceType.MultiResource
};
subRes = new List <ResourceInfo>();
}
long?size = null;
foreach (var subnode in downloads)
{
var href = subnode.GetAttributeValue("href", "");
if (href.IndexOf("ed2k://") == 0)
{
var reg = Regex.Match(href, "\\|file\\|([^\\|]+)\\|(\\d+)\\|([a-z\\d]+)", RegexOptions.IgnoreCase);
if (!reg.Success)
{
continue;
}
var filename = UD(reg.GetGroupValue(1)).Trim();
var hash = reg.GetGroupValue(3);
size = reg.GetGroupValue(2).ToInt64();
var res = CreateResourceInfo(hash, filename, ResourceType.Ed2K);
res.DownloadSizeValue = size;
//重置其它的Resource
if (parent != null)
{
parent.DownloadSizeValue = size;
}
subRes?.ForEach(s => s.DownloadSizeValue = size);
if (subRes != null)
{
subRes.Add(res);
}
else
{
result.Add(res);
}
}
else if (href.IndexOf("magnet:") == 0)
{
var reg = Regex.Match(href, "btih:([a-f\\d]{40})", RegexOptions.IgnoreCase);
if (!reg.Success)
{
continue;
}
var filename = UD(title).Trim();
var hash = reg.GetGroupValue(1);
var res = CreateResourceInfo(hash, filename, ResourceType.BitTorrent);
res.DownloadSize = sizestr;
res.DownloadSizeValue = size;
if (subRes != null)
{
subRes.Add(res);
}
else
{
result.Add(res);
}
}
else if (href.IndexOf("pan.baidu.com") != -1 || href.IndexOf("urlxf.qq.com") != -1 || href.IndexOf("kuai.xunlei.com") != -1 || href.IndexOf("yunpan.cn") != -1)
{
var type = href.IndexOf("pan.baidu.com") != -1 ? NetDiskType.BaiduDesk :
href.IndexOf("urlxf.qq.com") != -1 ? NetDiskType.QqXf :
href.IndexOf("kuai.xunlei.com") != -1 ? NetDiskType.XlShare : NetDiskType.QihuShare;
//百度网盘, 旋风分享， 迅雷快传
var res = CreateResourceInfo(null, title, ResourceType.NetDisk);
res.DownloadSize = sizestr;
res.DownloadSizeValue = size;
res.NetDiskData = new NetDiskData()
{
NetDiskType = type,
Url = href
};
if (subRes != null)
{
subRes.Add(res);
}
else
{
result.Add(res);
}
}
}
if (parent != null)
{
if (subRes.Count > 1)
{
parent.SubResources = subRes.Cast <IResourceInfo>().ToArray();
result.Add(parent);
}
else if (subRes.Count > 0)
{
result.Add(subRes[0]);
}
}
}
result.HasMore = context.Result.IndexOf(">下一页") != -1;
result.HasPrevious = result.PageIndex > 1;
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var document = CreateHtmlDocument(htmlContent);
var nodes = document.GetElementbyId("main")?.SelectNodes("ul/li");
if (nodes == null)
{
return(false);
}
foreach (var node in nodes)
{
var thumb = node.SelectSingleNode("div[@class='thumb']/a");
var titlelink = node.SelectSingleNode("div[@class='list_info']//h4/a");
if (titlelink == null)
{
continue;
}
var item = CreateResourceInfo(null, titlelink.GetAttributeValue("title", ""), ResourceType.MultiResource);
//<a href="/emule/download/2148521356" title="《权力的游戏 第五季》(Game of Thrones Season 5)[HDTVrip,WEBRip,720P,1080P]更新第9集">《<em class="highlight">权力的游戏</em> 第五季》(Game of Thrones Season 5)[HDTVrip,WEBRip,720P,1080P]更新第9集</a>
item.SiteData = new SiteInfo()
{
SiteID = Regex.Match(titlelink.GetAttributeValue("href", ""), @"\d+").Value.ToInt64()
};
//预览信息
if (thumb != null)
{
var style = thumb.GetAttributeValue("style", "");
if (style != null)
{
//background-image: url(http://thumb.icili.com/2148521356.jpg)
var pm = Regex.Match(style, @"url\(['""]?([^\)]+)['""]?\)");
if (pm.Success)
{
item.SupportPreivewType = PreviewType.Image;
item.PreviewInfo = new PreviewInfo()
{
ImageUrl = pm.GetGroupValue(1)
};
}
}
}
var introText = node.SelectSingleNode("div[@class='list_info']").InnerText;
//发布时间：2015-4-12 19:18 更新时间：2015-6-9 0:19
item.UpdateTimeDesc = Regex.Match(introText, @"发布时间.*?([\d-_]+\s*[\d-_:])").GetGroupValue(1);
// 文件数：44 | 点击数：2409
item.FileCount = Regex.Match(introText, @"文件数[\s：]*?([\d]+)").GetGroupValue(1).ToInt32Nullable();
item.HasSubResources = true;
item.ResourceType = ResourceType.MultiResource;
result.Add(item);
}
result.HasPrevious = result.PageIndex > 1;
var nextAnchor = document.DocumentNode.SelectSingleNode("//div[@class='pager']//a[last()]");
result.HasMore = nextAnchor != null && nextAnchor.GetAttributeValue("disabled", "") != "disabled";
return(base.LoadCore(context, url, htmlContent, result));
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
if (!base.LoadCore(context, url, htmlContent, result))
{
return(false);
}
var doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
var listTable = doc.DocumentNode.SelectSingleNode("//table");
if (listTable == null)
{
return(true);
}
var rows = listTable.SelectNodes("tbody/tr[not(contains(@class, 'success'))]");
var downloader = AppContext.Instance.DownloadServiceProviders.FirstOrDefault(s => s is MonovaDownloadProvider);
if (rows != null)
{
foreach (HtmlNode row in rows)
{
//var cls = row.GetAttributeValue("class", "");
//if (!cls.IsNullOrEmpty() && cls.IndexOf("success", StringComparison.OrdinalIgnoreCase) != -1)
// continue;
var namelink = row.SelectSingleNode("td[1]/a");
var link = namelink.Attributes["href"].Value;
var name = namelink.InnerText;
var item = new ResourceInfo()
{
Title = name,
Provider = this,
UpdateTimeDesc = row.SelectSingleNode("td[6]").InnerText,
DownloadSize = row.SelectSingleNode("td[5]").InnerText,
PreferDownloadProvider = downloader
};
var info = Regex.Match(link, @"torrent/(\d+)/([^""]*?)\.html", RegexOptions.IgnoreCase);
//可能是广告链接
if (!info.Success)
{
continue;
}
item.SiteData = new SiteInfo()
{
SiteID = info.Groups[1].Value.ToInt64(),
PageName = info.Groups[2].Value
};
result.Add(item);
}
}
//分页
result.HasPrevious = result.PageIndex > 1;
result.HasMore = (doc.DocumentNode.SelectSingleNode("//ul[contains(@class,'pagination')]/li[last()]")?.GetAttributeValue("class", "") ?? "active").IndexOf("active", StringComparison.OrdinalIgnoreCase) == -1;
return(base.LoadCore(context, url, htmlContent, result));
}