Exemplo n.º 1
0
        public List <DtoNewsUrlList> NewsUrlGathering(string newsListUrl, int newsType)
        {
            try
            {
                Log.Info(newsListUrl + " 抓取开始");
                var strContent = HttpHelper.GetContentByMobileAgent(newsListUrl, Encoding.GetEncoding("gb2312"));
                if (string.IsNullOrWhiteSpace(strContent))
                {
                    Log.Info(newsListUrl + " 未抓取到任何内容");
                    return(null);
                }

                //取得标题列表
                var strList = XpathHelper.GetInnerHtmlListByXPath(strContent, "//div[@class='leftList']/ul/li");

                if (strList != null && strList.Count > 0)
                {
                    foreach (var item in strList)
                    {
                        try
                        {
                            var url   = XpathHelper.GetAttrValueByXPath(item, "//a", "href");
                            var title = XpathHelper.GetInnerHtmlByXPath(item, "//a", "");
                            title = StrHelper.FormatHtml(title).Trim();


                            var isHave = DalNews.IsExistsNews(title);
                            //如果已存在则跳过
                            if (isHave)
                            {
                                continue;
                            }

                            if (newsType == 100 || newsType == 200 || newsType == 300)
                            {
                                #region === 根据详细页地址取新闻内容 ===

                                var news = NewsGathering(url);
                                if (news != null)
                                {
                                    news.NewsTypeId = newsType;
                                    news.Title      = title;

                                    news.PubTime = StrHelper.ToDateTime(StrHelper.FormatPubTime(news.PubTime.ToString()));
                                    //入库
                                    var newsId = DalNews.Insert(news);
                                    if (newsId < 1)
                                    {
                                        continue;
                                    }

                                    //从内容中提取img,存入newsmedia
                                    var mediaList = ImgDeal.GetImgList(news.Contents);
                                    if (mediaList != null && mediaList.Count > 0)
                                    {
                                        news.Contents = mediaList[0].Description;

                                        foreach (var picitem in mediaList)
                                        {
                                            picitem.NewsId = newsId;
                                            DalNews.InsertMedia(picitem);
                                        }
                                    }

                                    //休眠 控制抓取的频率
                                    Random rnd          = new Random();
                                    var    sleepSeconds = rnd.Next(30, 90);
                                    Thread.Sleep(sleepSeconds * 1000);
                                }

                                #endregion
                            }
                            if (newsType == 400)
                            {
                                #region === 根据详细页地址取图片内容 ===

                                var mediaList = NewsPicGathering(url);

                                var news = new DtoNews()
                                {
                                    Title      = title,
                                    FromUrl    = url,
                                    NewsTypeId = newsType,
                                };
                                if (mediaList != null && mediaList.Count > 0)
                                {
                                    news.Contents = mediaList[0].Description;

                                    //入库
                                    var newsId = DalNews.Insert(news);

                                    foreach (var picitem in mediaList)
                                    {
                                        picitem.NewsId = newsId;
                                        DalNews.InsertMedia(picitem);
                                    }
                                }

                                //休眠 控制抓取的频率
                                Random rnd          = new Random();
                                var    sleepSeconds = rnd.Next(30, 90);
                                Thread.Sleep(sleepSeconds * 1000);

                                #endregion
                            }
                            Log.Info(url + " 抓取完成");
                        }
                        catch (Exception ex)
                        {
                            //Log.Error("内容: " + item);
                            Log.Error(ex.Message + ex.StackTrace);
                        }
                    }
                }
                Log.Info(newsListUrl + " 抓取结束");
                return(null);
            }
            catch (Exception ex)
            {
                Log.Error(ex.Message + ex.StackTrace);
            }
            return(null);
        }