public ActionResult SingelSearch(string guid = null, string searchInfo = null)
{
CrawlerEntity crawlerEntity = null;
if (string.IsNullOrWhiteSpace(guid) && string.IsNullOrWhiteSpace(searchInfo))
{
ViewBag.Guid = Guid.NewGuid();
}
else
{
ViewBag.Guid = guid;
var count = new TargeCompanyDomain().Get(t => t.CompanyName.Equals(searchInfo))?.Count;
if (count > 0)
{
//历史记录中已存在
}
else
{
//上网检索
List<string> companyList = new List<string> { searchInfo };
TaskEntity model = new TaskEntity();
model.TaskType = EnumTaskType.BjCrawler;
model.TaskName = $"单个任务[{DateTime.Now.ToString("G")}]";
model.Unique = Conv.ToGuid(guid);
model.TaskStateDicId = 1;
model.TaskNum = 1;
model.CreateTime = DateTime.Now;
model.IsSingelSearch = true;
new TaskDomain().Add(model);
new BaseData(model).InsertMetadata(companyList.ToList(), model.TaskName, model, taskEntity =>
{
Task[] tasks = new Task[4];
for (int i = 0; i < 4; i++)
{
tasks[i] = new Task(() =>
{
var bjqyxy = new Crawler.Bjqyxy.BjCrawler(taskEntity, t => t.TaskGuid.Equals(taskEntity.Unique));
bjqyxy.SingelSearch(searchInfo);
});
tasks[i].Start();
}
Task.WaitAny(tasks);
});
}
crawlerEntity = new CrawlerDomain().Get(t => t.搜索名称 == searchInfo && t.名称 != null).FirstOrDefault();
}
return View(crawlerEntity);
}
public ActionResult CheckInsertMetadata(TaskEntity model)
{
var insertNum = new TargeCompanyDomain().GetInt(t => t.TaskGuid == model.Unique);
return Json(new { state = "doing", num = insertNum });
}
public ActionResult CheckGoGather(TaskEntity model)
{
var searchNum =
new TargeCompanyDomain().GetInt(
t => t.TaskGuid == model.Unique && t.IsSearched);
return Json(new { state = "doing", num = searchNum });
}
private void BaseWork()
{
bool isReloadCompany = true; //是否重新获取新的企业名称
string companyOld = ""; //记忆企业名称
var httpClient = new HttpClient();
httpClient.Setting.Timeout = 1000 * 5;
httpClient.Create<string>(HttpMethod.Post, firsturl).Send();
while (true)
{
var targetModel = new CrawlerEntity { 操作人姓名 = TaskEntity.OperatorName, 入爬行库时间 = TaskEntity.CreateTime, TaskGuid = TaskEntity.Unique };
try
{
//查询资源预处理
if (isReloadCompany)
{
var companyEntity = new TargeCompanyDomain().GetSingel(QueryCondition);
if (companyEntity == null)
break;
targetModel.搜索名称 = companyEntity.CompanyName; //搜索名称，直接持久化
companyOld = targetModel.搜索名称;
isReloadCompany = false;
}
else
{
targetModel.搜索名称 = companyOld;
}
//IP处理
var proxyEntity = new ProxyDomain().GetByRandom(); //代理IP
if (proxyEntity == null)
{
Console.WriteLine("在线代理临时获取策略启动。");
proxyEntity = Proxy.Proxy.GetInstance().GetHttProxyEntity();
Console.WriteLine("线上获取到了代理：{0}:{1}", proxyEntity.IpAddress, proxyEntity.Port);
}
httpClient.Setting.Proxy = new WebProxy(proxyEntity.IpAddress, proxyEntity.Port);
var resultBody = httpClient.Create<string>(HttpMethod.Post, targetUrl, data: new
{
queryStr = targetModel.搜索名称,
module = "",
idFlag = "qyxy"
}).Send();
if (!resultBody.IsValid())
{
RemoveOldIp(proxyEntity);
continue;
}
if (ValidText(resultBody.Result))
{
RemoveOldIp(proxyEntity);
continue;
}
//提取二级连接
var parser = new JumonyParser();
var urls = parser.Parse(resultBody.Result).Find("li a").ToList();
var nextUrl = "";
if (urls.Count < 1)
{
isReloadCompany = true;
AddNull(targetModel);
continue;
}
foreach (var htmlElement in urls)
{
targetModel.名称 = htmlElement.InnerText();
nextUrl = url + htmlElement.Attribute("href").AttributeValue;
}
//提取目标正文
var resultsecondBody =
httpClient.Create<string>(HttpMethod.Get, zhuUrl + new Uri(firsturl + nextUrl).Query).Send();
var nameValueCollection =
new NameValueCollection(URL.GetQueryString(new Uri(firsturl + nextUrl).Query));
if (!resultsecondBody.IsValid())
{
RemoveOldIp(proxyEntity);
continue;
}
if (ValidText(resultsecondBody.Result))
{
RemoveOldIp(proxyEntity);
continue;
}
//正文处理
var sorceIhtml = new JumonyParser().Parse(resultsecondBody.Result.Replace("<th", "<td"));
var tableLists = sorceIhtml.Find("table[class='f-lbiao']").ToList();
var listall = new List<string>();
foreach (var tableList in tableLists)
tableList.Find("tr td")
.ForEach(t => listall.Add(t.InnerText().TrimEnd(':').TrimEnd('：').Trim()));
var fillModel = FillModel(listall);
fillModel.全局唯一编号 = nameValueCollection["reg_bus_ent_id"].ToUpper();
new CrawlerDomain().Add(StrategyNo1(fillModel, targetModel));
//后续其他处理 包括了IP使用状态，以查询列表状态
proxyEntity.Usage = proxyEntity.Usage + 1;
new ProxyDomain().Update(proxyEntity);
Console.WriteLine("{0} 抓取到：{1}", Task.CurrentId, targetModel.搜索名称);
}
catch (Exception e)
{
isReloadCompany = true;
new LogDomain().Add(new LogEntity { LogType = "error", TaskName = TaskEntity.TaskName, ErrorDetails = Task.CurrentId + "线程： " + e.Message, Details = e.ToString(), TriggerTime = DateTime.Now });
continue;
}
isReloadCompany = true;
}
}