protected void OnButton1Clicked(object sender, EventArgs e)
{
if (!string.IsNullOrWhiteSpace(txtUrl.Text))
{
currentUrl = txtUrl.Text;
string htmlContent = HttpHelper.GetHttpContent(currentUrl);
Article article = HtmlToArticle.GetArticle(htmlContent);
if (article != null && !string.IsNullOrWhiteSpace(article.Title))
{
txtTitle.Text = article.Title;
txtPubDate.Text = article.PubDate.ToString();
if (!string.IsNullOrWhiteSpace(article.HtmlContent))
{
string htmlText = HtmlFormattor.FormatHtml(article.HtmlContent, currentUrl);
txtContent.Buffer.Text = htmlText;
}
}
else
{
txtTitle.Text = "";
txtPubDate.Text = "";
txtContent.Buffer.Text = "";
}
}
else
{
//MessageBox.Show("Please input your article url to extract.");
}
}
private static void WechatCrawler()
{
using (var client = new RedisAccess(redisPool))
{
try
{
//产生临时库（临时库和缓存总库取并集）
string wechatItem = client.PopItemWithHighestScoreFromSortedSet(wechatSubsKey);
if (!string.IsNullOrEmpty(wechatItem))
{
WeChatItemModel wechatItemModel = JsonConvert.DeserializeObject <WeChatItemModel>(wechatItem);
Palas.Protocol.PFItemToAnalyze pfItemToAnalyze = new Palas.Protocol.PFItemToAnalyze();
Palas.Protocol.PFItem pfItem = new Palas.Protocol.PFItem();
if (wechatItemModel != null && !string.IsNullOrEmpty(wechatItemModel.Id))
{
pfItem.AuthorID = wechatItemModel.WechatSubId;
pfItem.AuthorName = wechatItemModel.WechatSubName;
pfItem.ContentDetailLevel = Palas.Protocol.PFContentDetailLevel.FullContent;
pfItem.Crawler = "WechatSubs";
pfItem.CrawlID = wechatItemModel.WechatSubId;
pfItem.FetchTime = DateTimeUtility.GetUnixTimeStamp(DateTime.Now);
pfItem.ItemID = wechatItemModel.Id;
pfItem.MediaID = "WeChat";
pfItem.MediaName = "微信";
pfItem.PubDate = DateTimeUtility.GetUnixTimeStamp(wechatItemModel.PubDate);
pfItem.Url = wechatItemModel.Url.Replace("&", "&");
pfItem.CleanTitle = wechatItemModel.Title;
if (wechatItemModel.Tags != null && wechatItemModel.Tags.Count > 0)
{
pfItem.Tag = string.Join(",", wechatItemModel.Tags);
}
string content = GetHtmlContent(pfItem.Url);
if (!string.IsNullOrEmpty(content))
{
pfItem.HTMLText = HtmlFormattor.FormatHtml(content, pfItem.Url);
pfItem.CleanText = HTMLCleaner.CleanHTML(pfItem.HTMLText, true);
pfItemToAnalyze.AnalyzeFlag = 312287;
pfItemToAnalyze.Item = pfItem;
pfItemToAnalyze.CrawlRecode = new Palas.Protocol.PFCrawlRecode(pfItem.Crawler, pfItem.CrawlID, "Common", "", "N", 312287, 1, "WeChatSync", 0, 0, 0, "", 0, 0, true, pfItem.MediaID, pfItem.MediaName, pfItem.MediaChannel);
pfItemToAnalyze.CrawlRecode.Tags = pfItem.Tag;
pfItemToAnalyze.CrawlRecode.JoinIssueIDs = string.Join(",", wechatItemModel.Customers);
pfItemToAnalyze.CrawlRecode.MediaType = 11;
pfItemToAnalyze.CrawlRecode.MediaWeight = 8;
// 临时追加
if (wechatItemModel.Tags != null && wechatItemModel.Tags.Count > 0)
{
pfItem.Tag = string.Join(",", wechatItemModel.Tags);
//pfItemToAnalyze.CrawlRecode.Tags = pfItem.Tag;
}
// 临时追加完毕
//pfItemToAnalyze.CrawlRecode = new Palas.Protocol.PFCrawlRecode(pfItem.Crawler, pfItem.CrawlID, "Common", "", "N", 312287, string.Join(",", wechatItemModel.Customers), 1, "WeChatSync", 0, 0, "", 0, "", "", 0, 0, "", true, 4, pfItem.MediaID, pfItem.MediaName, "", 11, 0, 8, 0, "", "",);
Palas.Protocol.PFItemToAnalyze analyzeResult = AnalyzeItem.Analyzer(pfItemToAnalyze, Palas.Protocol.PFAnalyzeFlag.Splite);
//6.使用AnalyzeFirst对文章进行第一次分析
MultriAnalyzeFlag analyzeFlags = AnalyzeItem.BuildAnalyzeFlag((Enums.AnalyzeFlag)analyzeResult.AnalyzeFlag);
Palas.Protocol.PFItemToAnalyze analyzeFirstResult = AnalyzeItem.AnalyzerFirst(analyzeResult, analyzeFlags);
//7.使用AnalyzeSecond对文章进行第二次分析
Palas.Protocol.PFItemToAnalyze analyzeSecondResult = AnalyzeItem.AnalyzeSecond(analyzeFirstResult, analyzeFlags);
//8.使用AnalyzeIssue对文章进行分Issue分析
Palas.Protocol.PFItemToAnalyze analyzeIssueResult = AnalyzeItem.IssueAnalyzer(analyzeSecondResult, analyzeFlags);
//9.使用IndexThenDup将文章去重索引到ES
if (analyzeIssueResult != null)
{
//此处对Item进行一次转换
Palas.Protocol.PFItem pfItem2 = analyzeIssueResult.Item;
Item _item = TypeExchangeUtility.ExchangeItem(pfItem2);
_item = FilterIssue.FilterExcludeExpression(_item);
Enums.ProcessStatus result = Enums.ProcessStatus.Failed;
int retry = 0;
do
{
try
{
//retry++;
ESAccess.IndexOnly(_item);
result = Enums.ProcessStatus.Succeeded;
//result = DupThenIndexBusiness.DupThenIndexItem(_item);
//if (result == Enums.ProcessStatus.Failed)
//Thread.Sleep(new TimeSpan(0, 0, 30));
}
catch //(Exception ex)
{
result = Enums.ProcessStatus.Failed;
Thread.Sleep(new TimeSpan(0, 1, 00));
}
}while (result == Enums.ProcessStatus.Failed && retry < 3);
//10.判断是否成功
//Assert.AreNotEqual(Enums.ProcessStatus.Failed, result);
Console.WriteLine(string.Format("Index paper: {0}, Status: {1}", _item.ItemID, result.ToString()));
}
return;
}
}
}
else
{
//if have nothing to emit, then sleep for a little while to release CPU
Thread.Sleep(50);
}
}
catch (Exception ex)
{
}
}
}