public void AnalysisHtml(DataReceivedEventArgs args)
{
string htmlStr = args.Html;
var htmlDocument = new HtmlDocument();
htmlDocument.LoadHtml(htmlStr);
HtmlNode html = htmlDocument.DocumentNode;
var title = html.CssSelect("title");
if (title != null && title.Count() == 1)
{
Console.WriteLine(title.FirstOrDefault().InnerHtml);
Console.WriteLine(args.Url);
Console.WriteLine();
}
}
public bool IPLimitProcess(SimpleCrawler.DataReceivedEventArgs args)
{
try
{
if (string.IsNullOrEmpty(args.Html) || args.Html.Contains("503 Service Unavailable"))
{
return(true);
}
HtmlDocument document = new HtmlDocument();
document.LoadHtml(args.Html);
if (document.GetElementbyId("cy_center") == null)
{
Console.WriteLine("查找不到ID对象");
return(true);
}
}
catch (Exception)
{
}
return(false);
}
public void DataReceive(SimpleCrawler.DataReceivedEventArgs args)
{
try
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(args.Html);
string authorization = args.urlInfo.Authorization;
HtmlNode elementbyId = document.GetElementbyId("cy_center");
if (elementbyId == null)
{
Console.WriteLine("查找不到ID对象");
}
else
{
BsonDocument bsonDoc = new BsonDocument();
HtmlNode img = elementbyId.SelectSingleNode("./div/div[@class='cy-icon-box']/img");
if ((img != null) && (img.Attributes["src"] != null))
{
BsonDocument document3 = SaveProductImg(img, "projThumb");
bsonDoc.Set("projLocalThumb", document3.Text("localSrc"));
bsonDoc.Set("projThumb", document3.Text("src"));
}
HtmlNode node3 = elementbyId.SelectSingleNode("//h1[@class='cy-cp-name']");
if (node3 != null)
{
bsonDoc.Add("name", node3.InnerText.Trim());
}
HtmlNode node4 = elementbyId.SelectSingleNode("//div[@class='cy-xq-time']");
if (node4 != null)
{
bsonDoc.Add("pushDate", node4.InnerText.Replace("发布时间：", "").Trim());
}
HtmlNode node5 = elementbyId.SelectSingleNode("//div[@class='cy-tag-list']");
if (node5 != null)
{
bsonDoc.Add("tag", node5.InnerText.Trim());
}
HtmlNode node6 = elementbyId.SelectSingleNode("//div[@class='cy-cp-intro']");
if (node6 != null)
{
bsonDoc.Add("companyIntro", node6.InnerText.Trim());
}
HtmlNode node7 = elementbyId.SelectSingleNode("//div[@class='cy-cp-intro-info']");
if (node7 != null)
{
bsonDoc.Add("productIntro", node7.InnerText.Trim());
}
HtmlNode node8 = document.GetElementbyId("business");
if (node8 != null)
{
bsonDoc.Add("domain", node8.InnerText.Trim());
}
if (document.GetElementbyId("shareholders") != null)
{
bsonDoc.Add("shareholders", node8.InnerText);
}
HtmlNodeCollection imgs = elementbyId.SelectNodes("//ul[@class='gallery-img-box']/li/img");
if (imgs != null)
{
BsonDocument document4 = SaveProductImgs(imgs, bsonDoc.Text("pushDate"));
bsonDoc.Add("srcList", document4.Text("srcList"));
bsonDoc.Add("localSrcList", document4.Text("localSrcList"));
}
HtmlNode node10 = document.GetElementbyId("advantage");
if (node10 != null)
{
bsonDoc.Add("advantage", node10.InnerText.Trim());
}
HtmlNode node11 = document.GetElementbyId("results");
if (node11 != null)
{
bsonDoc.Add("achievements", node11.InnerText.Trim());
}
HtmlNodeCollection nodes2 = elementbyId.SelectNodes("//ul[@class='cy-cp-team']/li");
if (nodes2 != null)
{
BsonArray array = new BsonArray();
foreach (HtmlNode node14 in (IEnumerable <HtmlNode>)nodes2)
{
HtmlNode node15 = node14.SelectSingleNode("./div[@class='team-personnel-name']");
HtmlNode node16 = node14.SelectSingleNode("./div[@class='team-personnel-position']");
HtmlNode node17 = node14.SelectSingleNode("./div[@class='team-personnel-intro']");
if (((node15 != null) && (node16 != null)) && (node17 != null))
{
BsonDocument document5 = new BsonDocument {
{
"name",
node15.InnerText.Trim()
},
{
"position",
node16.InnerText.Trim()
},
{
"intro",
node17.InnerText.Trim()
}
};
array.Add(document5);
}
}
bsonDoc.Add("teamInfo", array);
}
HtmlNode node12 = elementbyId.SelectSingleNode("//div[@data-type='agree_chuangye']");
if (node12 != null)
{
bsonDoc.Add("agree", node12.InnerText.Trim());
}
HtmlNode node13 = elementbyId.SelectSingleNode("//div[@data-type='disagree_chuangye']");
if (node13 != null)
{
bsonDoc.Add("disAgree", node13.InnerText.Trim());
}
HtmlNodeCollection nodes3 = elementbyId.SelectNodes("//div[@class='box-moder cy-box-moder company-info']/ul/li");
if (nodes3 != null)
{
foreach (HtmlNode node18 in (IEnumerable <HtmlNode>)nodes3)
{
string[] separator = new string[] { "：", ":" };
string[] strArray = node18.InnerText.Trim().Split(separator, StringSplitOptions.RemoveEmptyEntries);
if (strArray.Length >= 2)
{
bsonDoc.Set(strArray[0].Trim(), strArray[1].Trim());
}
}
}
HtmlNodeCollection nodes4 = elementbyId.SelectNodes("//div[@class='box-moder cy-box-moder company-info get-company-box hide']/ul/li");
if (nodes4 != null)
{
foreach (HtmlNode node19 in (IEnumerable <HtmlNode>)nodes4)
{
string[] separator = new string[] { "：", ":" };
string[] strArray2 = node19.InnerText.Trim().Split(separator, StringSplitOptions.RemoveEmptyEntries);
if (strArray2.Length >= 2)
{
bsonDoc.Set(strArray2[0].Trim(), strArray2[1].Trim());
}
}
}
HtmlNodeCollection nodes5 = elementbyId.SelectNodes("//div[@class='box-moder cy-box-moder company-info get-company-box']/ul/li");
if (nodes5 != null)
{
foreach (HtmlNode node20 in (IEnumerable <HtmlNode>)nodes5)
{
string[] separator = new string[] { "：", ":" };
string[] strArray3 = node20.InnerText.Trim().Split(separator, StringSplitOptions.RemoveEmptyEntries);
if (strArray3.Length >= 2)
{
if (strArray3[0].Trim() == "办公地点")
{
bsonDoc.Set("地址", strArray3[1].Trim());
}
else
{
bsonDoc.Set(strArray3[0].Trim(), strArray3[1].Trim());
}
}
}
}
if (bsonDoc.ElementCount > 0)
{
bsonDoc.Set("isUpdate", "2");
Console.WriteLine(bsonDoc.Text("name") + "更新成功");
StorageData target = new StorageData
{
Document = bsonDoc,
Query = Query.EQ("_id", ObjectId.Parse(authorization)),
Name = this.DataTableName,
Type = StorageType.Update
};
DBChangeQueue.Instance.EnQueue(target);
}
if (UrlQueue.Instance.Count == 0)
{
Task.WaitAll(allTask.ToArray());
Task.WhenAll(allTask).GetAwaiter().OnCompleted(() => Console.WriteLine("所有下载完成"));
Console.ReadLine();
}
}
}
catch (Exception exception)
{
Console.WriteLine(exception.Message);
}
}
/// <summary>
/// The master data received event.
/// </summary>
/// <param name="args">
/// The args.
/// </param>
private void MasterDataReceivedEvent(DataReceivedEventArgs args)
{
_htmlAnalysisService.AnalysisHtml(args);
}
