Ejemplo n.º 1
0
        private void RunThread()
        {
            ScrapperMin sm = ScrapperMin.InitWC();

            sm.Log = new Action <string>(AddLog);
            string[] result = sm.Multiple(Scripts, DataList);
        }
Ejemplo n.º 2
0
        public void AddLink(string str)
        {
            InitAll();
            string[] ss = str.Split(new char[] { ' ' });
            if (ss == null || ss.Length == 0 || ss.Length != 2)
            {
                return;
            }
            string uu = AppConfig.GetThreadLink();

            Uri uri = null;

            try
            {
                uri = new Uri(uu);
            }
            catch
            {
                AddLog("[Ignore] IsThreadUri:false, url:'" + uu + "'");
                return;
            }

            string url = string.Format(uu, ss[0], ss[1]);

            if (ALL.Contains(url))
            {
                return;
            }

            lock (LoginLock)
            {
                if (HasLogin == false && AppConfig.GetRequireLogin())
                {
                    string scriptprefix = AppConfig.GetScriptPrefix();
                    if (File.Exists(Path.Combine(ScriptDir, scriptprefix + LoginScript)) == false)
                    {
                        return;
                    }
                    string[] strx = null;
                    try
                    {
                        strx = sm.Multiple(File.ReadAllText(Path.Combine(ScriptDir, scriptprefix + LoginScript)), new List <string> {
                            User, Pass
                        });
                    }
                    catch { }
                    if (strx != null && strx.Length == 1 && strx[0] != "SUCCESS LOGIN")
                    {
                        return;
                    }
                    HasLogin = true;
                }
            }

            string page = string.Empty;

            try
            {
                page = sm.WC.MethodPage(url, "GET", "", new Dictionary <string, string>());
            }
            catch { }
            if (string.IsNullOrEmpty(page))
            {
                return;
            }

            string title = "";
            string next  = AppConfig.GotNext();

            if (page.IndexOf(next) >= 0)
            {
                AddAllPage(url);
            }

            string[] tts = StringOps.TagMatch(page, "<title>", "</title>");
            if (tts != null && tts.Length > 0)
            {
                title = tts[0];
            }
            if (SetText != null)
            {
                SetText("[" + ss[1].ToString() + "] " + title);
            }

            string regex      = AppConfig.GetRegex();
            var    linkParser = new Regex(regex, RegexOptions.Compiled | RegexOptions.IgnoreCase);
            var    matchesCol = linkParser.Matches(page);

            if (matchesCol == null || matchesCol.Count == 0)
            {
                AddLog("[NotFound] title:'" + title + "', url:'" + url + "'");
                return;
            }

            List <string> domains = new List <string>();

            string[] doms    = AppConfig.GetDomains().Split(new char[] { '\n' });
            string[] ignores = AppConfig.GetIgnoreDomains().Split(new char[] { '\n' });
            if (ignores.Length > 0 && ignores[0] == "")
            {
                ignores = new string[0];
            }
            string[] bodies = AppConfig.GetIgnoreBodies().Split(new char[] { '\n' });
            if (bodies.Length > 0 && bodies[0] == "")
            {
                bodies = new string[0];
            }
            string[] bodiesMatches = AppConfig.GetBodies().Split(new char[] { '\n' });
            if (bodies.Length > 0 && bodies[0] == "")
            {
                bodies = new string[0];
            }

            if (doms.Length > 0 && doms[0] != "")
            {
                foreach (string dom in doms)
                {
                    domains.Add(dom.Trim());
                }
            }

            if (bodies.Length > 0 && bodies[0] != "")
            {
                foreach (string body in bodies)
                {
                    if (page.ToLower().Contains(body.ToLower()))
                    {
                        AddLog("[Ignore] ContainsBody:'" + body + "', title:'" + title + "', url:'" + url + "'");
                        return;
                    }
                }
            }
            if (bodiesMatches.Length > 0 && bodiesMatches[0] != "")
            {
                bool contains = false;
                foreach (string body in bodiesMatches)
                {
                    if (body.ToLower().StartsWith("http") && page.ToLower().Contains(body.ToLower()))
                    {
                        contains = true;
                        break;
                    }
                    else if (body.ToLower().StartsWith("http") == false)
                    {
                        if (page.ToLower().Contains(body.ToLower()))
                        {
                            contains = true;
                            break;
                        }
                    }
                }
                if (contains == false)
                {
                    AddLog("[Ignore] NotContainsBody:true, title:'" + title + "', url:'" + url + "'");
                    return;
                }
            }

            int count = 0;

            foreach (Match m in matchesCol)
            {
                if (m == null || string.IsNullOrEmpty(m.Value))
                {
                    continue;
                }

                string urlx = m.Value;
                if (urlx.Contains("imgur"))
                {
                    Console.WriteLine("a");
                }
                if (urlx.Contains("\""))
                {
                    urlx = m.Value.Substring(0, m.Value.IndexOf("\""));
                }
                if (urlx.Contains("<"))
                {
                    urlx = m.Value.Substring(0, m.Value.IndexOf("<"));
                }
                if (urlx.Contains(".") == false)
                {
                    AddLog("[Ignore] ContainsDot:false, title:'" + title + "', url:'" + url + "', urlx:'" + urlx + "'");
                    continue;
                }
                if (urlx.Contains("..."))
                {
                    AddLog("[Ignore] Contains3Dots:true, title:'" + title + "', url:'" + url + "', urlx:'" + urlx + "'");
                    continue;
                }
                try
                {
                    Uri urlxx = new Uri(urlx);
                }
                catch
                {
                    AddLog("[Ignore] IsUri:false, title:'" + title + "', url:'" + url + "', urlx:'" + urlx + "'");
                    continue;
                }
                bool ig = false;
                foreach (string ig2 in ignores)
                {
                    if (urlx.ToLower().Contains(ig2.ToLower()))
                    {
                        ig = true;
                        AddLog("[Ignore] ig2:'" + ig2 + "', title:'" + title + "', url:'" + url + "', urlx:'" + urlx + "'");
                        break;
                    }
                }
                if (ig)
                {
                    continue;
                }

                if (domains.Count > 0)
                {
                    foreach (string dom in domains)
                    {
                        if (urlx.ToLower().IndexOf(dom.ToLower()) < 0)
                        {
                            continue;
                        }

                        count++;
                        AddFile(title, url, urlx);
                        break;
                    }
                }
                else
                {
                    count++;
                    AddFile(title, url, urlx);
                }
            }
            if (count == 0)
            {
                AddLog("[NotFound] title:'" + title + "', url:'" + url + "'");
            }
        }