public static void CrawlerAnswer(int qid)
{
var proxys = GrabAnswer.GetProxyListFromCache();
//var answer = new QuestionParser().ParseAnswer(item.QuestionId.ToString());
var answerHttpClient = CrawlerSingleQuestion(qid, proxys);
if (answerHttpClient == null || string.IsNullOrWhiteSpace(answerHttpClient.ToString()))
{
return;
}
var xd = answerHttpClient[0]["questions"][0]["xd"].ToString();
var chid = answerHttpClient[0]["questions"][0]["chid"].ToString();
var apiUrl = $"https://www.zujuan.com/question/list?question_id={qid}&xd={xd}&chid={chid}";
var apiJson = HttpClientHolder.Execute(apiUrl);
using (var db = new CrawlerEntities())
{
var entity = new QuestionAll();
entity.QuestionId = qid;
entity.IsDelete = false;
entity.AnswerJson = answerHttpClient.ToString();
entity.ApiJson = apiJson;
entity.CrawlerUrl = $"https://www.zujuan.com/question/detail-{qid}.shtml";
entity.CrawlerApiUrl = apiUrl;
entity.child = chid.NullToInt();
entity.xd = xd.NullToInt();
db.QuestionAll.Add(entity);
db.SaveChanges();
}
}
public static void CrawlerAnswer(QuestionAll item, List <string> proxys)
{
//var answer = new QuestionParser().ParseAnswer(item.QuestionId.ToString());
var answerHttpClient = CrawlerSingleQuestion(item.QuestionId.ToString(), proxys);
if (answerHttpClient == null || string.IsNullOrWhiteSpace(answerHttpClient.ToString()))
{
//Console.WriteLine($"anserhttp is null {item.QuestionId}");
//UpdateQuestionGrabStatus(item.QuestionId);
return;
}
var xd = answerHttpClient[0]["questions"][0]["xd"].ToString();
var chid = answerHttpClient[0]["questions"][0]["chid"].ToString();
var apiUrl = $"https://www.zujuan.com/question/list?question_id={item.QuestionId}&xd={xd}&chid={chid}";
var apiJson = HttpClientHolder.Execute(apiUrl, GetCookieState(chid.NullToInt(), xd.NullToInt()));
Console.WriteLine($"save sql");
UpdateQuesion(item.QuestionId, answerHttpClient, apiUrl, apiJson, xd, chid);
}
public static void StartWithIndex()
{
int qid = CacheManager.IncrementValue(key).NullToInt();
if (qid > 10000000)
{
System.Windows.Forms.Application.Exit();
return;
}
if (!IsCrawlered(qid))
{
var result = HttpClientHolder.Execute($"https://www.zujuan.com/question/detail-{qid}.shtml");
if (result.IndexOf("试题已经被删除") >= 0)
{
AddDeleteQuestion(qid);
}
else
{
CrawlerAnswer(qid);
}
}
}
/// <summary>
/// 整站采集，如果整体数据过一遍，需要重新生成这些category下每页的URL、
/// </summary>
public void InitCatePageUrl()
{
var listCategory = DataService.GetCategorylist();
var hasCateID = DataService.GetCrawleredCatelist();
//https://www.zujuan.com/question?categories=47854&bookversion=47832&nianji=47854&chid=3&xd=1
foreach (var category in listCategory)
//Parallel.ForEach(listCategory, new ParallelOptions() { MaxDegreeOfParallelism = 10 }, (category) =>
{
if (hasCateID.Any(t => t == category.CategoryId))
{
continue;
}
Stopwatch sw = new Stopwatch();
sw.Start();
//通过页面采集方式，放弃
//var url = $"https://www.zujuan.com/question?categories={category.CategoryId}&bookversion={category.BookVersionId}&nianji={category.CategoryId}&chid={category.Child}&xd={category.Degree}";
//var html = new HttpUnitHelperNosingleton().GetRealHtmlTrice(url);
//var doc = NSoupClient.Parse(html);
//var total = doc.Select("div.total b")[0].Text().NullToInt();
//DataService.UpdateCategoryCount(category.Id, total);
//var pageNum = total / 10 + 1;
//for (int i = 1; i <= pageNum; i++)
//{
// var grabUrl = "";
// if (i > 1)
// {
// grabUrl =
// $"https://www.zujuan.com/question/index?categories={category.CategoryId}&bookversion={category.BookVersionId}&nianji=7741&chid={category.Child}&xd={category.Degree}&page={i}&per-page=10";
// }
// else
// {
// grabUrl = url;
// }
// DataService.AddCateUrl(grabUrl, category.CategoryId, i);
//}
var api =
$"https://www.zujuan.com/question/list?categories={category.CategoryId}&sortField=time&page=1&_=1540841532659";
var url = $"https://www.zujuan.com/question?categories={category.CategoryId}&bookversion={category.BookVersionId}&nianji={category.CategoryId}&chid={category.Child}&xd={category.Degree}";
var html = new HttpUnitHelper().GetRealHtmlOnceNotWaitJs(url);
var questionCount = NSoupClient.Parse(html).GetElementById("J_QuestionList").GetElementsByTag("li").Count;
long start = 1000000000000;
long end = 1540849999999;
long randomId = 1540840000000 + new Random().Next(0000000, 9999999);
//todo 题型筛选
var cookies = new HttpUnitHelper().webClient.GetCookies(new URL("https://www.zujuan.com/"));
var headerCookie = string.Empty;
foreach (var cookie in cookies)
{
headerCookie += $"{cookie.Name}={cookie.Value};";
}
var json = HttpClientHolder.Execute(api, headerCookie);
var total = JObject.Parse(json);
var totalCount = total["total"].NullToInt();
if (totalCount <= 0)
{
continue;
}
var pageNum = totalCount / 10 + 1;
Parallel.For(1, pageNum, (i) =>
{
var currentApi =
$"https://www.zujuan.com/question/list?categories={category.CategoryId}&sortField=time&page={i}&_={randomId}";
var currentJson = HttpClientHolder.Execute(currentApi, headerCookie);
var grabUrl = "";
if (i > 1)
{
grabUrl =
$"https://www.zujuan.com/question/index?categories={category.CategoryId}&bookversion={category.BookVersionId}&nianji=7741&chid={category.Child}&xd={category.Degree}&page={i}&per-page=10";
}
else
{
grabUrl = url;
}
Action actoin = () =>
{
DataService.AddCateUrl(grabUrl, category.CategoryId, i, currentApi, currentJson);
};
actoin.BeginInvoke(null, null);
});
sw.Stop();
Debug.WriteLine("cost" + sw.ElapsedMilliseconds);
}
//);
}
