ln.skyscanner/crawl/Crawler.cs

// /**
//  * File: Crawler.cs
//  * Author: haraldwolff
//  *
//  * This file and it's content is copyrighted by the Author and / or copyright holder.
//  * Any use wihtout proper permission is illegal and may lead to legal actions.
//  *
//  *
//  **/
using System;
using ln.types.threads;
using System.Net;
using System.Collections.Generic;
using System.IO;
using ln.logging;
using ln.types;
using System.Linq;
using ln.types.serialize;
using ln.skyscanner.entities;
using System.Net.NetworkInformation;
using ln.snmp;
using ln.snmp.endpoint;
using ln.perfdb;
using ln.perfdb.storage;
using ln.skyscanner.check;
using System.Threading;
using ln.snmp.types;
using Newtonsoft.Json;
using Newtonsoft.Json.Converters;
using ln.types.odb;
using ln.skyscanner.crawl.service;
using ln.skyscanner.crawl.tests;
using LiteDB;

namespace ln.skyscanner.crawl
{
    public class Crawler
    {
        static Crawler()
        {
            CrawlService.RegisterService(new TCP(new int[] { 13080, 13022, 80, 22, 443, 13443 }));
            CrawlService.RegisterService(new SNMP(new string[] { "VhclfC7lfIojYZ", "Vhclf(C7$lfIojYZ", "ByFR4oW98hap", "qVy3hnZJ2fov" }));
            CrawlService.RegisterService(new RFC1213());
        }

        public SkyScanner SkyScanner { get; }

        public String BasePath { get; set; }
        public String PerfPath => Path.Combine(BasePath, "perfdb");
        public string PoolPath => Path.Combine(BasePath, "pool");
        public String DBFileName => Path.Combine(PoolPath, "crawler.db");

        bool stopping;

        Pool crawlThreadPool = new Pool(0);

        public PoolJob[] CurrentJobs => crawlThreadPool.CurrentPoolJobs;
        public PoolJob[] QueuedJobs => crawlThreadPool.QueuedJobs;

        public bool CrawlSubnets { get; set; }
        public bool CrawlHosts { get; set; }

        [JsonConverter(typeof(StringEnumConverter))]
        public ComponentState CrawlerState { get; private set; }

        Thread threadScheduler;

        public ODBCollection<CrawledHost> CrawledHosts => hosts;
        public ODBCollection<CrawledSubnet> CrawledSubnets => subnets;

        //private ODB<CrawlPool> dbCrawlPool;

        private ODB odbDatabase;
        private ODBCollection<CrawledHost> hosts;
        private ODBCollection<CrawledSubnet> subnets;


        public Crawler(SkyScanner skyScanner)
        {
            SkyScanner = skyScanner;
            try
            {
                BasePath = Path.Combine(skyScanner.BasePath, "crawler");

                if (!Directory.Exists(BasePath))
                    Directory.CreateDirectory(BasePath);
                if (!Directory.Exists(PerfPath))
                    Directory.CreateDirectory(PerfPath);
                if (!Directory.Exists(PoolPath))
                    Directory.CreateDirectory(PoolPath);

                //dbCrawlPool = new ODB<CrawlPool>(PoolPath);

                odbDatabase = new ODB(BasePath);
                hosts = odbDatabase.GetCollection<CrawledHost>();
                subnets = odbDatabase.GetCollection<CrawledSubnet>();

                CrawlerState = ComponentState.INITIALIZED;
            }
            catch (Exception)
            {
                Stop();

                throw;
            }
        }

        public void Start()
        {
            if (CrawlerState != ComponentState.STOPPING)
            {
                stopping = false;

                //if (dbCrawlPool == null)
                    //dbCrawlPool = new ODB<CrawlPool>(PoolPath);

                if (crawlThreadPool == null)
                    crawlThreadPool = new Pool(64);
                else
                    crawlThreadPool.SetPoolSize(64);

                if (threadScheduler == null)
                {
                    threadScheduler = new Thread(scheduler);
                    threadScheduler.Start();
                }

                CrawlerState = ComponentState.STARTED;
            }
        }

        public void Stop()
        {
            try
            {
                if (CrawlerState == ComponentState.STOPPING)
                {
                    if (crawlThreadPool != null)
                    {
                        crawlThreadPool.Abort();


                        if (crawlThreadPool.CurrentPoolSize == 0)
                            crawlThreadPool = null;
                    }
                }
                else
                {
                    CrawlerState = ComponentState.STOPPING;

                    stopping = true;
                    if (threadScheduler != null)
                    {
                        threadScheduler.Join();
                        threadScheduler = null;
                    }

                    crawlThreadPool.Close();

                    stopping = false;
                    CrawlerState = ComponentState.STOPPED;
                }
            }
            catch (Exception e)
            {
                Logging.Log(e);
                CrawlerState = ComponentState.FAILED;
            }
        }

        public void EnsureSubnet(CIDR network)
        {
            FindSubnet(network);
        }

        public void Enqueue(JobDelegate job)
        {
            crawlThreadPool.Enqueue(job);
        }

        public void Crawl(Guid hostID)
        {
            CrawledHost crawledHost = CrawledHosts[hostID];
            Crawl(crawledHost);
        }
        public void Crawl(CrawledHost crawledHost)
        {
            Crawl crawl = new Crawl(this, crawledHost);
            crawlThreadPool.Enqueue(crawl);
        }

        public void Crawl(CrawledSubnet subnet)
        {
            CrawlNetwork crawlSubnet = new CrawlNetwork(this, subnet);
            crawlThreadPool.Enqueue(crawlSubnet);
        }

        public CrawledHost FindHostForIP(CIDR ip)
        {
            CrawledHost crawledHost = CrawledHosts.Where(host => host.HasIP(ip)).FirstOrDefault();
            if (crawledHost == null)
            {
                crawledHost = new CrawledHost();
                crawledHost.PrimaryIP = ip;
                crawledHost.Name = ip.ToString();

                CrawledHosts.Insert(crawledHost);
            }
            return crawledHost;
        }
        public CrawledSubnet FindSubnet(CIDR network)
        {
            CrawledSubnet sn = CrawledSubnets.Where(subnet => subnet.Network.Equals(network)).FirstOrDefault();
            if (sn == null)
            {
                sn = new CrawledSubnet(network);
                CrawledSubnets.Insert(sn);
            }
            return sn;
        }


        private void scheduler()
        {
            int count = 0;

            while (!stopping)
            {
                count++;

                SkyScanner.Entities.GlobalNetwork.Update();

                if (CrawlHosts)
                    foreach (CrawledHost crawledHost in CrawledHosts.Where( host => (host.NextCheck < DateTime.Now)))
                    {
                        Crawl(crawledHost);
                    }

                if (CrawlSubnets)
                    foreach (CrawledSubnet subnet in CrawledSubnets.Where( sn => (sn.NextScan < (DateTime.Now - TimeSpan.FromDays(1)) && sn.Network.MaskWidth >= 24)))
                    {
                        Crawl(subnet);
                    }

                Thread.Sleep(5000);
            }
        }


    }
}