public void Crawler()
{
var listSubjectGrade = GetSubjectGrade();
var listArea = GetAreas();
while (ExistCrawlerJC())
{
var listJiaocai = GetRandom10CrawlerJC();
foreach (var jc in listJiaocai)
{
var grades = listSubjectGrade.Where(t => t.SubjectId == jc.SubjectId);
Parallel.ForEach(grades, (grade) =>
{
Parallel.ForEach(listArea, (area) =>
{
if (ExistGrabPageSource(area.AreaId, jc.JiaoCaiDetailId, grade.GradeId))
{
return;
}
var url =
$"http://zujuan.xkw.com/{jc.Prefix}/zj{jc.JiaoCaiDetailId}/a{area.AreaId}g{grade.GradeId}/";
Console.WriteLine(url);
var html = HttpWebResponseProxyMayi.ExecuteCreateGetHttpResponseProxy(url, 3000, null);
var totalCount = doc.GetElementById("questioncount").Text().NullToInt();
var pageCount = totalCount / 10 + 1;
AddGrabPageSource(area.AreaId, jc.JiaoCaiDetailId, grade.GradeId, totalCount);
//for (int i = 1; i <= pageCount; i++)
//{
// AddGrabPageList(area.AreaId, jc.JiaoCaiDetailId, grade.GradeId, totalCount, i);
//}
});
});
UpdateJiaocaiDetailStatus(jc.JiaoCaiDetailId);
}
}
}
public static JArray CrawlerSingleQuestion(string questionId)
{
try
{
#if DEBUG
var a =
HttpWebResponseProxyMayi.ExecuteCreateGetHttpResponseProxy(
$"https://www.zujuan.com/question/detail-{questionId}.shtml", 6000, null);
//var a = HttpClientHolder.Proxy_GetRequest2();
#else
var a = HttpClientHolder.Proxy_GetRequestAbyyun($"https://www.zujuan.com/question/detail-{questionId}.shtml");
#endif
//
if (a.IndexOf("试题已经被删除") >= 0)
{
Console.WriteLine($"{questionId} has delete!!!");
using (var db = new CrawlerEntities())
{
var id = questionId.NullToInt();
var entity = db.Question.FirstOrDefault(t => t.Question_Id == id);
if (entity != null)
{
entity.IsGrabAns = true;
entity.IsRemoteDelete = true;
db.SaveChanges();
}
}
return(null);
}
if (a.IndexOf("限制访问试题") >= 0)
{
using (var db = new CrawlerEntities())
{
var id = questionId.NullToInt();
var entity = db.Question.FirstOrDefault(t => t.Question_Id == id);
if (entity != null)
{
entity.IsGrabAns = false;
db.SaveChanges();
}
}
return(null);
}
Console.WriteLine($"start crawler https://www.zujuan.com/question/detail-{questionId}.shtml");
//例如我想提取记录中的NAME值
string value = GetValue(a, "var MockDataTestPaper =", "OT2.renderQList").TrimEnd(new char[] { ';' });
value = value.Trim().TrimEnd(new char[] { ';' }).Trim();
JArray jObject = JArray.Parse(value);
if (!string.IsNullOrWhiteSpace(jObject[0]["questions"][0]["list"].NullToString()))
{
if (string.IsNullOrWhiteSpace(jObject[0]["questions"][0]["list"][0]["answer"].NullToString()))
{
Console.WriteLine("明细题答案抓取错误");
Debug.WriteLine("明细题答案抓取错误");
throw new Exception("明细题答案抓取错误");
}
}
//; UpdateProxGrabyime(proxy.Id);
Console.WriteLine($"aleady get {questionId} return value");
//更新代理时间
return(JArray.Parse(value));
}
catch (IOException io)
{
using (var db = new CrawlerEntities())
{
var id = questionId.NullToInt();
var entity = db.Question.FirstOrDefault(t => t.Question_Id == id);
if (entity != null)
{
entity.IsRemoteDelete = false;
entity.IsGrabAns = false;
db.SaveChanges();
}
}
}
catch (JsonReaderException je)
{
using (var db = new CrawlerEntities())
{
var id = questionId.NullToInt();
var entity = db.Question.FirstOrDefault(t => t.Question_Id == id);
if (entity != null)
{
entity.IsRemoteDelete = false;
entity.IsGrabAns = false;
db.SaveChanges();
}
}
}
catch (WebException we)
{
using (var db = new CrawlerEntities())
{
var id = questionId.NullToInt();
var entity = db.Question.FirstOrDefault(t => t.Question_Id == id);
if (entity != null)
{
entity.IsRemoteDelete = false;
entity.IsGrabAns = false;
db.SaveChanges();
}
}
}
catch (Exception e)
{
using (var db = new CrawlerEntities())
{
var id = questionId.NullToInt();
var entity = db.Question.FirstOrDefault(t => t.Question_Id == id);
if (entity != null)
{
entity.IsRemoteDelete = false;
entity.IsGrabAns = false;
db.SaveChanges();
}
}
}
//UpdateQuestionGrabStatus(questionId.NullToInt());
return(null);
}