Files
A69WD/Src/A69WD/Net/cA69Engine3.cs
T
2022-09-22 11:29:19 +02:00

643 lines
28 KiB
C#

using System;
using System.IO;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Windows.Forms;
using System.Security.Policy;
using System.Net;
using static System.Windows.Forms.LinkLabel;
using System.ComponentModel;
using static System.Net.Mime.MediaTypeNames;
using System.Collections;
using System.Text.RegularExpressions;
namespace A69WD {
public class cA69Engine3 {
#region Property: State, NumDownload, Tipologie, BASE_URL, BASE_LINK_IDX, BASE_LINK_RAC
public EngineStatusEnum State { get { return state; } }
public int NumDownload { get { return downloaded.Count; } }
static string[] Tipologie = { "altro", "bdsm", "gay", "incesto", "lesbo", "orge", "prime", "scambio", "tradimenti", "trans", "trio" };
const string BASE_URL = "https://www.annunci69.it/racconti-erotici/{Tip}.html";
const string BASE_LINK_IDX = "https://www.annunci69.it/racconti-erotici/racconti_lista.php";
const string BASE_LINK_RAC = "https://www.annunci69.it/racconti-erotici/";
#endregion
#region Fields: allLinks, toDownload, downloaded, racLinks, dtStartTime, downRac, state
List<string> allLinks = null;
List<string> toDownload = null;
List<string> downloaded = null;
List<string> racLinks = null;
DateTime dtStartTime;
bool downRac = false;
EngineStatusEnum state = EngineStatusEnum.NotReady;
#endregion
#region Costruttore e distruttore
public cA69Engine3() {
try {
// Inizializzo gli archivi (dai file in memoria)
allLinks = new List<string>();
toDownload = new List<string>();
downloaded = new List<string>();
racLinks = new List<string>();
int res = App.DB.LoadData(allLinks, toDownload, downloaded, racLinks);
RaiseEvent_Updating("Ricaricati " + res.ToString("000") + " link di racconti da scaricare");
// Per ora non serializzo gli archivi e quindi carico le pagine iniziali delle 11 tipologie:
foreach (string tipologia in Tipologie) {
toDownload.Add(BASE_URL.Replace("{Tip}", tipologia));
allLinks.Add(BASE_URL.Replace("{Tip}", tipologia));
}
} catch (Exception ex) {
Logger.Log(new object[] { "Error Code: ER02220", "Errore in cA69Engine()", ex });
throw ex;
}
state = EngineStatusEnum.StandBy;
}
#endregion
#region Method:
public bool Start(bool DownRac) {
if (state != EngineStatusEnum.StandBy) return false;
dtStartTime = DateTime.Now;
downRac = DownRac;
RaiseEvent_Updating("Inizio lo scrap dei link dei racconti");
state = EngineStatusEnum.Starting;
if (toDownload.Count > 0) {
// proseguo con il download:
string lk = toDownload.First();
toDownload.Remove(lk);
string fn = Path.Combine(Cfg.Settings.PathAnalizadedRaccFile, GetFileName(lk));
using (MyWebClient client = new MyWebClient(lk, fn)) {
//client.Encoding = Encoding.ASCII;
client.Headers.Add("user-agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:88.0) Gecko/20100101 Firefox/88.0");
client.Headers.Add("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8");
client.Headers.Add("Cookie", "euconsent-v2=CPG74goPG74goAKAeAENBcCsAP_AAH_AAAwIH3tf_X__bX9j-_59__t0eY1f9_7_v-wzjhfdt-8N2P_X_L0X42E7PF36pq4KuR4Eu3LBIQNlHOHUTUmw6okVrTPsak2Mr7NKJ7LEmnMZe2dYGHtfn91TuZKY7_78_9fz3_-v_v___9f3r-3_3__59X---_e_V399zLv9__3__9gfiASYal8AF2JY4Mk0aVQogQhWEh0AoAKKAYWiawgYXBTsrAI9QQMAEJqAjAiBBiCjFgEAAgEASERASAHggEQBEAgABACpAQgAImAQWAFgYBAAKAaFiBFAEIEhBkcFRymBARItFBPZWAJRd7GmEIZRYAUCj-iowEShBAsDISFg5gAA.YAAAAAAAAAAA");
client.Headers.Add("DNT", "1");
client.Headers.Add("TE", "Trailers");
//client.DownloadFile(new Uri(lk), fn);
client.DownloadProgressChanged += Client_DownloadProgressChanged;
client.DownloadFileCompleted += Client_DownloadFileCompleted;
client.DownloadDataCompleted += Client_DownloadDataCompleted;
client.DownloadStringCompleted += Client_DownloadStringCompleted;
//client.DownloadFileAsync(new Uri(lk), fn);
//client.DownloadDataAsync(new Uri(lk));
client.DownloadStringAsync(new Uri(lk));
}
}
return true;
}
public bool Stop() {
switch (state) {
case EngineStatusEnum.NotReady:
case EngineStatusEnum.StandBy:
SaveState();
state = EngineStatusEnum.Stopped;
return true;
case EngineStatusEnum.Starting:
case EngineStatusEnum.Started:
case EngineStatusEnum.Stopping:
state = EngineStatusEnum.Stopping;
return false;
case EngineStatusEnum.Stopped:
return true;
}
return true;
}
public int ImportLinkFromFile(string SrcFilePath) {
int res = 0;
try {
if (File.Exists(SrcFilePath)) {
StreamReader sr = new StreamReader(SrcFilePath, Encoding.UTF8);
string row = sr.ReadLine().Trim();
while (row != null) {
// verifico che non sia un commento:
if (!(row.StartsWith("#") || row.StartsWith("//"))) {
// Se link nuovo:
if (!allLinks.Contains(row)) {
// Link nuovo, mi comporto di conseguenza alla tipologia:
if (IsRacLink(row)) {
// Link ad un racconto
allLinks.Add(row);
toDownload.Add(row);
racLinks.Add(row);
res++;
} else {
if (IsIdxLink(row)) {
// Link ad un indice
allLinks.Add(row);
toDownload.Add(row);
res++;
} else {
// Link non riconosciuto, lo scarto
;
}
}
}
}
//Read the next line
row = sr.ReadLine();
}
}
} catch (Exception ex) {
Logger.Log(new object[] { "Error Code: DB07403", "Errore in ImportLinkFromFile()", ex });
if (res > 0) res = -1 * res;
}
return res;
}
#endregion
#region Eventi:
/// <summary>
/// Evento che indica un cambiamento di stato nell'AnalizeManager
/// </summary>
public event EventHandler<EngineEventArgs> Updating;
#region Generazione degli eventi dell'engine verso le classi chiamanti:
/// <summary>
/// Attiva l'evento che segnala un cambiamento di stato nel engine (per esempio, terminato il download di un singolo file)
/// </summary>
void RaiseEvent_Updating(string msg, int tot, int toDown, int down) {
EventHandler<EngineEventArgs> tmp = Updating;
if (null != tmp) tmp(this, new EngineEventArgs(msg, tot, toDown, down));
}
/// <summary>
/// Attiva l'evento che segnala un cambiamento di stato nel engine (per esempio, terminato il download di un singolo file)
/// </summary>
void RaiseEvent_Updating(string msg) {
EventHandler<EngineEventArgs> tmp = Updating;
if (null != tmp) tmp(this, new EngineEventArgs(msg, allLinks.Count, toDownload.Count, downloaded.Count));
}
#endregion
#endregion
int conteggioEvt = 0;
private void Wb_DocumentCompleted(object sender, WebBrowserDocumentCompletedEventArgs e) {
object obj = new Object();
lock (obj) {
conteggioEvt++;
StringBuilder sb;
if (state == EngineStatusEnum.Starting) state = EngineStatusEnum.Started;
WebBrowser wb = sender as WebBrowser;
string link = wb.Url.ToString();
// Se link di logon, devo terminare i download
if (link == "https://www.annunci69.it/accedi.php?nop") {
// Finito download:
state = EngineStatusEnum.Stopping;
// salvo gli archivi:
SaveState();
return;
}
// Elimino ulteriori ripetizioni dell'evento:
if (downloaded.Contains(link)) return;
// aggiorno UI
RaiseEvent_Updating(wb.Name + " - (Evt" + conteggioEvt.ToString("0000") + ") Donwloaded: " + link.Replace("https://www.annunci69.it/racconti-erotici/", "./"));
// Se pagina è un racconto, lo salvo
if (IsRacLink(link)) {
sb = new StringBuilder();
sb.Append(wb.Document.Body.InnerHtml);
string fn = Path.Combine(Cfg.Settings.PathAnalizadedRaccFile, GetFileName(link));
//cTools.WriteTextFile(fn, sb, Encoding.GetEncoding(wb.Document.Encoding));
cTools.WriteTextFile(fn, sb);
}
// Registro che ho eseguito il download del link
downloaded.Add(link);
// Ciclo su tutti i link estratti dalla pagina:
foreach (HtmlElement item in wb.Document.Links) {
link = item.GetAttribute("href");
// Se link nuovo:
if (!allLinks.Contains(link)) {
// elimino il link di logon
if (link == "https://www.annunci69.it/accedi.php?nop") break;
// Link nuovo, mi comporto di conseguenza alla tipologia:
if (IsRacLink(link)) {
// Link ad un racconto
allLinks.Add(link);
if (downRac) { toDownload.Add(link); }
racLinks.Add(link);
} else {
if (IsIdxLink(link)) {
// Link ad un indice
allLinks.Add(link);
toDownload.Add(link);
} else {
// Link non riconosciuto, lo scarto
;
}
}
}
} // fine ciclo sui link estratti
// Se ho raggiunto il massimo numero di link, o non vi sono link da downloadare, mi fermo:
if ((toDownload.Count > 0) && (state == EngineStatusEnum.Started)) {
// proseguo con il download:
string lk = toDownload.First();
toDownload.Remove(lk);
wb.Navigate(lk);
} else {
// Finito download:
state = EngineStatusEnum.Stopping;
// salvo gli archivi:
SaveState();
}
}//lock
}
private void SaveState() {
int res = App.DB.SaveData(allLinks, toDownload, downloaded, racLinks);
StringBuilder sb;
// Finito download, salvo gli archivi:
string dt = DateTime.Now.ToString("yyyy-MM-dd_HH-mm-ss");
// Salvo archivio: AllLinks
sb = new StringBuilder();
allLinks.Sort();
foreach (string lnk in allLinks) { sb.AppendLine(lnk); }
cTools.WriteTextFile(Path.Combine(Cfg.Settings.PathAnalizadedIndexFile, string.Format("_A69_AllLinks_{0}.tes", dt)), sb);
// Salvo archivio: toDownload
sb = new StringBuilder();
toDownload.Sort();
foreach (string lnk in toDownload) { sb.AppendLine(lnk); }
cTools.WriteTextFile(Path.Combine(Cfg.Settings.PathAnalizadedIndexFile, string.Format("_A69_toDownload_{0}.tes", dt)), sb);
// Salvo archivio: Downloaded
sb = new StringBuilder();
downloaded.Sort();
foreach (string lnk in downloaded) { sb.AppendLine(lnk); }
cTools.WriteTextFile(Path.Combine(Cfg.Settings.PathAnalizadedIndexFile, string.Format("_A69_Downloaded_{0}.tes", dt)), sb);
// Salvo archivio: RacLinks
sb = new StringBuilder();
racLinks.Sort();
foreach (string lnk in racLinks) { sb.AppendLine(lnk); }
cTools.WriteTextFile(Path.Combine(Cfg.Settings.PathAnalizadedIndexFile, string.Format("_A69_RacLinks_{0}.tes", dt)), sb);
// avviso utente:
RaiseEvent_Updating(string.Format("Finito il download in {0}", DateTime.Now - dtStartTime));
state = EngineStatusEnum.Stopped;
}
private string GetFileName(string url) {
if (IsRacLink(url)) {
return url.Replace("https://www.annunci69.it/racconti-erotici/", "A69_").Replace("/", "_").Replace(":", "_").Replace(@"\", "_");
} else {
return url.Replace("https://www.annunci69.it/racconti-erotici/", "_A69_").Replace("/", "_").Replace(":", "_").Replace(@"\", "_");
}
}
static private bool IsIdxLink(string link) { return link.StartsWith(BASE_LINK_IDX); }
static private bool IsRacLink(string link) { foreach (string tip in Tipologie) { if (link.StartsWith(BASE_LINK_RAC + tip + "/")) return true; } return false; }
static public string LinkType(string link) {
if (IsIdxLink(link)) return "Idx";
if (IsRacLink(link)) return "Rac";
//return "Unknown";
return "Unk";
}
private void Client_DownloadFileCompleted(object sender, AsyncCompletedEventArgs e) {
object obj = new Object();
lock (obj) {
conteggioEvt++;
StringBuilder sb;
if (state == EngineStatusEnum.Starting) state = EngineStatusEnum.Started;
WebClient wb = sender as WebClient;
string link = wb.BaseAddress;
return;
/*
// Se link di logon, devo terminare i download
if (link == "https://www.annunci69.it/accedi.php?nop") {
// Finito download:
state = EngineStatusEnum.Stopping;
// salvo gli archivi:
SaveState();
return;
}
// Elimino ulteriori ripetizioni dell'evento:
if (downloaded.Contains(link)) return;
// aggiorno UI
RaiseEvent_Updating(wb.Name + " - (Evt" + conteggioEvt.ToString("0000") + ") Donwloaded: " + link.Replace("https://www.annunci69.it/racconti-erotici/", "./"));
// Se pagina è un racconto, lo salvo
if (IsRacLink(link)) {
sb = new StringBuilder();
sb.Append(wb.Document.Body.InnerHtml);
string fn = Path.Combine(Cfg.Settings.PathAnalizadedRaccFile, GetFileName(link));
//cTools.WriteTextFile(fn, sb, Encoding.GetEncoding(wb.Document.Encoding));
cTools.WriteTextFile(fn, sb);
}
// Registro che ho eseguito il download del link
downloaded.Add(link);
// Ciclo su tutti i link estratti dalla pagina:
foreach (HtmlElement item in wb.Document.Links) {
link = item.GetAttribute("href");
// Se link nuovo:
if (!allLinks.Contains(link)) {
// elimino il link di logon
if (link == "https://www.annunci69.it/accedi.php?nop") break;
// Link nuovo, mi comporto di conseguenza alla tipologia:
if (IsRacLink(link)) {
// Link ad un racconto
allLinks.Add(link);
if (downRac) { toDownload.Add(link); }
racLinks.Add(link);
} else {
if (IsIdxLink(link)) {
// Link ad un indice
allLinks.Add(link);
toDownload.Add(link);
} else {
// Link non riconosciuto, lo scarto
;
}
}
}
} // fine ciclo sui link estratti
// Se ho raggiunto il massimo numero di link, o non vi sono link da downloadare, mi fermo:
if ((toDownload.Count > 0) && (state == EngineStatusEnum.Started)) {
// proseguo con il download:
string lk = toDownload.First();
toDownload.Remove(lk);
wb.Navigate(lk);
} else {
// Finito download:
state = EngineStatusEnum.Stopping;
// salvo gli archivi:
SaveState();
}
//*/
}//lock
}
private void Client_DownloadDataCompleted(object sender, DownloadDataCompletedEventArgs e) {
//System.Threading.AutoResetEvent waiter = (System.Threading.AutoResetEvent)e.UserState;
try {
// If the request was not canceled and did not throw
// an exception, display the resource.
if (!e.Cancelled && e.Error == null) {
byte[] data = (byte[])e.Result;
string textData = System.Text.Encoding.Unicode.GetString(data);
string FullFileName = @"C:\VMS\Prj\C#\A69WD\Test\Txt\Src\2022-09-22\A69_altroz.html";
//System.IO.StreamWriter sw = new System.IO.StreamWriter(FullFileName, false, Encoding.Unicode);
//sw.Write(textData);
//sw.Flush();
//sw.Close();
using (var fs = new FileStream(FullFileName, FileMode.Create, FileAccess.Write)) {
fs.Write(data, 0, data.Length);
}
}
} finally {
// Let the main application thread resume.
//waiter.Set();
}
}
private void Client_DownloadStringCompleted(object sender, DownloadStringCompletedEventArgs e) {
object obj = new Object();
lock (obj) {
conteggioEvt++;
MyWebClient mwb = sender as MyWebClient;
if (state == EngineStatusEnum.Starting) state = EngineStatusEnum.Started;
// Verifico se download corretto
if (!e.Cancelled && e.Error == null) {
// aggiorno UI
RaiseEvent_Updating("(Evt" + conteggioEvt.ToString("0000") + ") Donwloaded: " + mwb.URL.Replace("https://www.annunci69.it/racconti-erotici/", "./"));
// Se pagina è un racconto, lo salvo
if (IsRacLink(mwb.URL)) {
string textData = e.Result;
string FullFileName = mwb.FileName;
System.IO.StreamWriter sw = new System.IO.StreamWriter(FullFileName, false, Encoding.UTF8);
sw.Write(textData);
sw.Flush();
sw.Close();
}
// Registro che ho eseguito il download del link
downloaded.Add(mwb.URL);
// Estraggo tutti i link net testo:
List<string> links = new List<string>();
//if (EstraiA69Link(e.Result, links)) {
if (EstraiLink(e.Result, "/racconti-erotici/", links)) {
// Ciclo su tutti i link estratti dalla pagina:
foreach (string link in links) {
if (!allLinks.Contains(link)) {
// elimino il link di logon
if (link == "https://www.annunci69.it/accedi.php?nop") break;
// Link nuovo, mi comporto di conseguenza alla tipologia:
if (IsRacLink(link)) {
// Link ad un racconto
allLinks.Add(link);
if (downRac) { toDownload.Add(link); }
racLinks.Add(link);
} else {
if (IsIdxLink(link)) {
// Link ad un indice
allLinks.Add(link);
toDownload.Add(link);
} else {
// Link non riconosciuto, lo scarto
;
}
}
}
}
} // fine ciclo sui link estratti
} else {
// Download con errore, informo l'utente
RaiseEvent_Updating("(Evt" + conteggioEvt.ToString("0000") + ") Donwload error: " + mwb.URL.Replace("https://www.annunci69.it/racconti-erotici/", "./"));
}
// Se ho raggiunto il massimo numero di link, o non vi sono link da downloadare, mi fermo:
if ((toDownload.Count > 0) && (state == EngineStatusEnum.Started)) {
// proseguo con il download:
string lk = toDownload.First();
toDownload.Remove(lk);
string fn = Path.Combine(Cfg.Settings.PathAnalizadedRaccFile, GetFileName(lk));
using (MyWebClient client = new MyWebClient(lk, fn)) {
//client.Encoding = Encoding.ASCII;
client.Headers.Add("user-agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:88.0) Gecko/20100101 Firefox/88.0");
client.Headers.Add("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8");
client.Headers.Add("Cookie", "euconsent-v2=CPG74goPG74goAKAeAENBcCsAP_AAH_AAAwIH3tf_X__bX9j-_59__t0eY1f9_7_v-wzjhfdt-8N2P_X_L0X42E7PF36pq4KuR4Eu3LBIQNlHOHUTUmw6okVrTPsak2Mr7NKJ7LEmnMZe2dYGHtfn91TuZKY7_78_9fz3_-v_v___9f3r-3_3__59X---_e_V399zLv9__3__9gfiASYal8AF2JY4Mk0aVQogQhWEh0AoAKKAYWiawgYXBTsrAI9QQMAEJqAjAiBBiCjFgEAAgEASERASAHggEQBEAgABACpAQgAImAQWAFgYBAAKAaFiBFAEIEhBkcFRymBARItFBPZWAJRd7GmEIZRYAUCj-iowEShBAsDISFg5gAA.YAAAAAAAAAAA");
client.Headers.Add("DNT", "1");
client.Headers.Add("TE", "Trailers");
//client.DownloadFile(new Uri(lk), fn);
client.DownloadProgressChanged += Client_DownloadProgressChanged;
client.DownloadFileCompleted += Client_DownloadFileCompleted;
client.DownloadDataCompleted += Client_DownloadDataCompleted;
client.DownloadStringCompleted += Client_DownloadStringCompleted;
//client.DownloadFileAsync(new Uri(lk), fn);
//client.DownloadDataAsync(new Uri(lk));
client.DownloadStringAsync(new Uri(lk));
}
} else {
// Finito download:
state = EngineStatusEnum.Stopping;
// salvo gli archivi:
SaveState();
}
}//lock
}
private void Client_DownloadProgressChanged(object sender, DownloadProgressChangedEventArgs e) {
////Update progress bar & label
//Invoke(new MethodInvoker(delegate () {
// progressBar.Minimum = 0;
// double receive = double.Parse(e.BytesReceived.ToString());
// double total = double.Parse(e.TotalBytesToReceive.ToString());
// double percentage = receive / total * 100;
// lblStatus.Text = $"Downloaded {string.Format("{0:0.##}", percentage)}%";
// progressBar.Value = int.Parse(Math.Truncate(percentage).ToString());
//}));
}
public static bool EstraiA69Link(string txt, List<string> links) {
try {
Regex regex = new Regex("(?<=href=\")\\S+?(?=\")");
Match matchObj = regex.Match(txt);
while (matchObj.Success) {
if (matchObj.Value.StartsWith("/racconti-erotici/")) { links.Add(matchObj.Value); }
if (matchObj.Groups.Count > 1) {
for (int i = 1; i < matchObj.Groups.Count; i++) {
Group group = matchObj.Groups[i];
if (group.Success) {
if (group.Value.StartsWith("/racconti-erotici/")) { links.Add(group.Value); }
}
}
}
matchObj = matchObj.NextMatch();
}
return true;
} catch { return false; }
}
public static bool EstraiLink(string txt, string startWith, List<string> links) {
try {
Regex regex = new Regex("(?<=href=\")\\S+?(?=\")");
Match matchObj = regex.Match(txt);
while (matchObj.Success) {
if (matchObj.Value.StartsWith(startWith)) { links.Add(matchObj.Value.Replace(startWith, "https://www.annunci69.it/racconti-erotici/")); }
if (matchObj.Groups.Count > 1) {
for (int i = 1; i < matchObj.Groups.Count; i++) {
Group group = matchObj.Groups[i];
if (group.Success) {
if (group.Value.StartsWith(startWith)) { links.Add(group.Value.Replace(startWith, "https://www.annunci69.it/racconti-erotici/")); }
}
}
}
matchObj = matchObj.NextMatch();
}
return true;
} catch { return false; }
}
}//class cA69Engine3
public class MyWebClient : WebClient {
public string URL;
public string FileName;
public MyWebClient(string url, string fileName) {
this.URL = url;
this.FileName = fileName;
}
}
}//namespace A69WD