public PageActor(string webSocket, XpathConfigurationForPuppeteer configuration)
{
PageActorName = Self.Path.Name;
WebSocket = webSocket;
Configuration = configuration;
ActorTaskScheduler.RunTask(async() =>
{
await PreparePage();
});
}
public BrowserActor(XpathConfigurationForPuppeteer xpathConfiguration)
{
XpathConfiguration = xpathConfiguration;
BrowserActorName = Self.Path.Name;
AmountOfPages = 3;
// TODO Add Chromium instance handling
// TODO Add Chromium instance handling
// TODO Add Chromium instance handling
// TODO Add Chromium instance handling
// TODO Add Chromium instance handling
}
public ProjectActor(ProjectDefinition projectDefinition)
{
ProjectActorName = Self.Path.Name;
List <ScraperSchema> scraperSchemas = projectDefinition.ScraperSchemas;
XpathConfigurationForPuppeteer crawlerConfiguration = new XpathConfigurationForPuppeteer(projectDefinition);
Context.ActorOf(Props.Create <UrlParserActor>(), ActorPaths.UrlParser.Name); // TODO Add coordinator for scaling
Context.ActorOf(Props.Create <ObjectParserActor>(scraperSchemas), ActorPaths.ObjectParser.Name); // TODO Add coordinator for scaling
Context.ActorOf(Props.Create <BrowserActor>(crawlerConfiguration), ActorPaths.Browser.Name); // TODO Add coordinator for scaling
#region SingleTons
IActorRef urlTracker = Context.ActorOf(Props.Create <UrlTrackerActor>(projectDefinition), ActorPaths.UrlTracker.Name); // there can NEVER be more than ONE instance!!!!
Context.ActorOf(Props.Create <ObjectTrackerActor>(scraperSchemas), ActorPaths.ObjectTracker.Name); // there can NEVER be more than ONE instance!!!!
#endregion
urlTracker.Tell(new UnprocessedUrlsMessage(projectDefinition.StartUrls)); // START URL
}
