public static MovieInfo GetMovieInfoByOnlineURL(string onlineURL, bool isContainIntro = false)
{
var html = HTTPHelper.GetHTMLByURL(onlineURL);
if (string.IsNullOrEmpty(html))
{
return(null);
}
var htmlDom = htmlParser.Parse(html);
var nameDom = htmlDom.QuerySelector("h1.font14");
var introDom = htmlDom.QuerySelector("div.Drama_c");
var infoTable = htmlDom.QuerySelectorAll("tr.CommonListCell");
var pubDate = DateTime.Now;
if (infoTable != null && infoTable.Length > 2)
{
if (infoTable[1] != null && !string.IsNullOrEmpty(infoTable[1].TextContent))
{
DateTime.TryParse(infoTable[1].TextContent.Replace("发布时间", "").Replace("\n", ""), out pubDate);
}
}
return(new MovieInfo()
{
MovieName = nameDom != null ? nameDom.InnerHtml : "获取名称失败...",
Dy2018OnlineUrl = onlineURL,
MovieIntro = introDom != null && isContainIntro ? introDom.InnerHtml : "",
PubDate = pubDate,
});
}
public static void CrawlHotClickMovieInfo(int endIndex = 10)
{
Task.Factory.StartNew(() =>
{
try
{
LogHelper.Info("CrawlHotClickMovieInfo Start...");
for (int index = 1; index <= endIndex; index++)
{
var indexURL = $"http://www.btdytt520.com/hotclick/p_{index}.html";
var html = HTTPHelper.GetHTMLByURL(indexURL);
if (string.IsNullOrEmpty(html))
{
return;
}
var htmlDom = htmlParser.Parse(html);
foreach (var li in htmlDom.QuerySelectorAll("li.newsli"))
{
var aDom = li.QuerySelectorAll("a").FirstOrDefault(a => !string.IsNullOrEmpty(a.GetAttribute("target")));
if (aDom == null)
{
continue;
}
var onlineURL = "http://www.btdytt520.com/" + aDom.GetAttribute("href");
if (hotMoviceHelper.IsContainsMoive(onlineURL) || li.QuerySelector("li.phlidate") == null)
{
continue;
}
var pubDate = DateTime.Now;
DateTime.TryParse(li.QuerySelector("li.phlidate").InnerHtml, out pubDate);
hotMoviceHelper.AddToMovieDic(new MovieInfo()
{
Dy2018OnlineUrl = onlineURL,
MovieName = aDom.InnerHtml,
PubDate = pubDate,
});
}
}
LogHelper.Info("CrawlHotClickMovieInfo Finish.");
}
catch (Exception ex)
{
LogHelper.Error("CrawlHotClickMovieInfo Exception", ex);
LogHelper.Info("CrawlHotClickMovieInfo Finish.");
}
});
}
public static void CrawlLatestMovieInfo()
{
var indexURL = "http://www.btdytt520.com/movie/";
var html = HTTPHelper.GetHTMLByURL(indexURL);
if (string.IsNullOrEmpty(html))
{
return;
}
var htmlDom = htmlParser.Parse(html);
var divMovie = htmlDom.QuerySelector("div.index_Sidebar_cc");
var lstMovie = divMovie.QuerySelectorAll("a").Select(a => new MovieInfo()
{
Dy2018OnlineUrl = "http://www.btdytt520.com/" + a.GetAttribute("href"),
MovieName = a.InnerHtml
}).ToList();
}
/// <summary>
/// 爬取数据
/// </summary>
public static void CrawlHotMovie()
{
Task.Factory.StartNew(() =>
{
try
{
LogHelper.Info("CrawlHotMovie Start...");
var htmlDoc = HTTPHelper.GetHTMLByURL("http://www.dy2018.com/");
var dom = htmlParser.Parse(htmlDoc);
var lstDivInfo = dom.QuerySelectorAll("div.co_content222");
if (lstDivInfo != null)
{
//前三个DIV为新电影
foreach (var divInfo in lstDivInfo.Take(3))
{
divInfo.QuerySelectorAll("a").Where(a => a.GetAttribute("href").Contains("/i/")).ToList().ForEach(
a =>
{
var onlineURL = "http://www.dy2018.com" + a.GetAttribute("href");
if (!hotMovieList.IsContainsMoive(onlineURL))
{
MovieInfo movieInfo = Dy2018MoviceInfoHelper.GetMovieInfoFromOnlineURL(onlineURL);
if (movieInfo != null && movieInfo.XunLeiDownLoadURLList != null && movieInfo.XunLeiDownLoadURLList.Count != 0)
{
hotMovieList.AddToMovieDic(movieInfo);
}
}
});
}
}
LogHelper.Info("CrawlHotMovie Finish...");
}
catch (Exception ex)
{
LogHelper.Error("CrawlHotMovie Exception", ex);
}
});
}
public static void CrawlHostMovieInfo()
{
var indexURL = "http://www.btdytt520.com/movie/";
var html = HTTPHelper.GetHTMLByURL(indexURL);
if (string.IsNullOrEmpty(html))
{
return;
}
var htmlDom = htmlParser.Parse(html);
var divMovie = htmlDom.QuerySelector("div.index_Sidebar_cc");
divMovie.QuerySelectorAll("a").Select(a => a).ToList().ForEach(
a =>
{
var aURL = "http://www.btdytt520.com" + a.GetAttribute("href");
if (!hotMoviceHelper.IsContainsMoive(aURL))
{
hotMoviceHelper.AddToMovieDic(Btdytt520Helper.GetMovieInfoByOnlineURL(aURL));
}
});
}
/// <summary>
/// 从在线网页提取电影数据
/// </summary>
/// <param name="onlineURL"></param>
/// <returns></returns>
public static MovieInfo GetMovieInfoFromOnlineURL(string onlineURL, bool isContainIntro = false)
{
try
{
var movieHTML = HTTPHelper.GetHTMLByURL(onlineURL);
if (string.IsNullOrEmpty(movieHTML))
{
return(null);
}
var movieDoc = htmlParser.Parse(movieHTML);
var zoom = movieDoc.GetElementById("Zoom");
var lstDownLoadURL = movieDoc.QuerySelectorAll("[bgcolor='#fdfddf']");
var updatetime = movieDoc.QuerySelector("span.updatetime"); var pubDate = DateTime.Now;
if (updatetime != null && !string.IsNullOrEmpty(updatetime.InnerHtml))
{
DateTime.TryParse(updatetime.InnerHtml.Replace("发布时间：", ""), out pubDate);
}
var lstOnlineURL = lstDownLoadURL.Select(a => a.QuerySelector("a")).Where(item => item != null).Select(item => item.InnerHtml).ToList();
var movieName = movieDoc.QuerySelector("div.title_all");
var movieInfo = new MovieInfo()
{
MovieName = movieName != null && movieName.QuerySelector("h1") != null?
movieName.QuerySelector("h1").InnerHtml : "找不到影片信息...",
Dy2018OnlineUrl = onlineURL,
MovieIntro = zoom != null && isContainIntro?WebUtility.HtmlEncode(zoom.InnerHtml) : "暂无介绍...",
XunLeiDownLoadURLList = lstOnlineURL,
PubDate = pubDate,
};
return(movieInfo);
}
catch (Exception ex)
{
LogHelper.Error("GetMovieInfoFromOnlineURL Exception", ex, new { OnloneURL = onlineURL });
return(null);
}
}
/// <summary>
/// 从在线网页提取数据
/// </summary>
/// <param name="i"></param>
private static void FillMovieFromOnline(string indexURL)
{
var htmlDoc = HTTPHelper.GetHTMLByURL(indexURL);
var dom = htmlParser.Parse(htmlDoc);
var lstDivInfo = dom.QuerySelectorAll("div.co_content8");
if (lstDivInfo != null)
{
lstDivInfo.FirstOrDefault().QuerySelectorAll("a").Where(a => a.GetAttribute("href").Contains("/i/")).ToList()
.ForEach(a =>
{
var onlineURL = "http://www.dy2018.com" + a.GetAttribute("href");
if (!latestMovieList.IsContainsMoive(onlineURL))
{
MovieInfo movieInfo = Dy2018MoviceInfoHelper.GetMovieInfoFromOnlineURL(onlineURL);
if (movieInfo != null && movieInfo.XunLeiDownLoadURLList != null && movieInfo.XunLeiDownLoadURLList.Count != 0)
{
latestMovieList.AddToMovieDic(movieInfo);
}
}
});
}
}