public void CrawlStepperToDatabase(int Id)
{
var stepperEng = new eqranews.crawling.Models.CrawlStepper();
List <eqranews.crawling.Models.CrawlResult> results = new List <eqranews.crawling.Models.CrawlResult>();
using (var scope = _serviceProvider.CreateScope())
{
// var services = _serviceProvider;
ApplicationDbContext db = scope.ServiceProvider.GetRequiredService <ApplicationDbContext>();
if (db.CrawlSteppers.Where(x => x.Id == Id).Any())
{
DAL.Crawling.CrawlStepper stepper = db.CrawlSteppers.Include(s => s.CrawlSteps).ThenInclude(i => i.CrawlItems).Include("CrawlSteps.CrawlStepType").Where(x => x.Id == Id).SingleOrDefault();
foreach (DAL.Crawling.CrawlStep step in stepper.CrawlSteps.ToList())
{
List <eqranews.crawling.Models.CrawlItem> CrawlItems = new List <eqranews.crawling.Models.CrawlItem>();
var newParams = new object[] { step.Id, crawling.Models.CrawlSteps.CrawlSetpType.Single, new AngleSharp.Url(step.Url), step.Selector };
step.CrawlItems.ToList().ForEach(item =>
{
CrawlItems.Add(new eqranews.crawling.Models.CrawlItem {
Name = item.Name, Selector = item.Selector, Attr = item.Attr, Value = item.Value
});
});
string StepClassFullName = String.Format("eqranews.crawling.Models.CrawlSteps.{0}, eqranews.crawling, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null", step.CrawlStepType.Name);
Type t = Type.GetType(StepClassFullName);
if (t == null)
{
return;
}
var instance = (eqranews.crawling.Models.CrawlSteps.CrawlStep)Activator.CreateInstance(t, newParams);
var myPropInfo = t.GetProperty("CrawlItems");
myPropInfo.SetValue(instance, CrawlItems);
// Add step to stepper
stepperEng.StepSequence.Add(instance);
}
;
results = stepperEng.Crawl();
SaveToStore(results, db, stepper);
}
}
}
// Save the results in the database
public void SaveToStore(List <CrawlResult> results, ApplicationDbContext db, DAL.Crawling.CrawlStepper stepper)
{
try
{
foreach (CrawlResult item in results)
{
if (!db.News.Any(N => N.SourceLink == item.Url.Href))
{
var source = db.CrawlSources.Find(new object[] { stepper.CrawlSourceId });
News newsEntry = new News
{
Title = item.CrawlItems.Where(x => x.Name == "Title").FirstOrDefault()?.Value,
SourceLink = item.Url.Href,
Created = DateTime.Now,
SourceId = source.Id,
CountryId = source.CountryId
};
db.News.Add(newsEntry);
db.SaveChanges();
if (newsEntry.Id > 0)
{
// Add the Main Category to the News Item Category
if ((int)stepper.CategoryId > 0)
{
db.NewsCategories.Add(new NewsCategory()
{
Id = 0, NewsId = newsEntry.Id, CategoryId = (int)stepper.CategoryId, Main = true
});
db.SaveChanges();
}
// Add all the News Items
foreach (var x in item.CrawlItems)
{
NewsItem nItem = new NewsItem
{
NewsId = newsEntry.Id,
Name = x.Name,
Value = x.Value
};
if (nItem.Name == "Category")
{
int cat = 0;
if (int.TryParse(nItem.Value, out cat))
{
db.NewsCategories.Add(new NewsCategory()
{
Id = 0, NewsId = newsEntry.Id, CategoryId = cat, Main = false
});
db.SaveChanges();
}
}
else
{
db.NewsItems.Add(nItem);
}
}
db.SaveChanges();
}
}
}
}
catch (Exception)
{
}
}
