Участник:TohaomgBot

Материал c сайта НП «Викимедиа РУ»
Перейти к навигации Перейти к поиску

Этот бот заменяет все вхождения определённого растрового изображения на его аналог в формате SVG, например "какое-то_изображение.jpg" на "какое-то_изображение.svg". Пара изображений подбирается вручную. Код программы можно увидеть внизу (Я разрешаю кому угодно использовать, распространять и изменять его, но не получать от него финансовую прибыль). Бот написан на языке C# с использованием библиотеки DotNetWikiBot.

Код

using System;
using System.Net;
using System.Threading;
using System.Collections.Generic;
using System.Text.RegularExpressions;
using DotNetWikiBot;

class ImageReplacer:Bot
{
 [STAThread]
 public static void Main(string[] args)
 {
  string username = args[0], password = args[1];
  string new_img;
  List<string> sites_list = new List<string>();
  Dictionary<string, string> summary = new Dictionary<string, string>();
  int edits, rejects;
  MatchCollection mc;
  
  summary.Add("en", "BOT: Replaced raster image with an image of format SVG.");
  summary.Add("de", "BOT: Ersetzt das Rasterbild mit einem Bild des Formats SVG.");
  summary.Add("pt", "BOT: Substituição de imagem raster por formato SVG.");
  summary.Add("uk", "БОТ: Замінено растрове зображення на зображення в форматі SVG.");
  summary.Add("ru", "БОТ: Заменено растровое изображение на изображение в формате SVG.");
  summary.Add("be", "БОТ: Заменены растравы малюнак на малюнак у фармаце SVG.");
  summary.Add("pl", "BOT: Zamień grafikę rastrową na obraz w formacie SVG.");
  summary.Add("sr", "BOT: Zamenjena rasterska slika sa slikom u formatu SVG.");
  summary.Add("hr", "BOT: Zamijenjena rasterska slika slikom u formatu SVG.");
  summary.Add("hu", "BOT: Kép(ek) cseréje SVG formátumú változatra.");
  summary.Add("ar", "BOT: استبدال صورة نقطية بصورة متجهية متغيرة (SVG).");
  summary.Add("fa", "BOT: جایگزینی پرونده با نسخهٔ برداری SVG");
  summary.Add("nv", "BÉÉSH HINÁANII: Eʼelyaaígíí łahgo áyiilaa; SVG choolʼį́.");
  
  beginning:
  edits = 0;


  //USER INPUT
  Console.WriteLine("Replace: ");
  string old_img = args.Length>=3?args[2]:Console.ReadLine();
	if(args.Length>=3) {Console.WriteLine(old_img);}
  if(old_img.StartsWith("id="))
  {
   Site commons_site = new Site("https://commons.wikimedia.org/", username, password);
   Page commons_page = new Page(commons_site, Int64.Parse(old_img.Substring(3)));
   commons_page.Load();
   old_img = commons_page.title.Substring(5);
   Console.WriteLine(old_img);
 
   mc = Regex.Matches(commons_page.text, @"\|((.)+svg)");
   new_img = mc[0].Groups[1].Value;
   
   new_img = new_img.Replace("File:", "");
   if(new_img.Contains("|")) {new_img = new_img.Substring(new_img.IndexOf("|")+1);}
   if(new_img.Contains("_")) {new_img = new_img.Replace("_", " ");}
 
   Console.WriteLine("\n"+"with: \n" + new_img);
  }
  else
  {
   Console.WriteLine("\n"+"with: ");
   new_img = args.Length>=4?args[3]:Console.ReadLine();
   	if(args.Length>=4) {Console.WriteLine(new_img);}
   if(new_img==old_img)
   {Console.WriteLine("\nNew image can not be the same as the old one!\n"); goto beginning;}
  }
  string old_img_underlined = old_img.Replace(" ", "_");
  Console.WriteLine();
 

  //WHAT WIKIS TO PROCESS
  string htmlCode;
  string img_usage_list = "https://" + "commons.wikimedia.org/w/index.php?title=Special:GlobalUsage&limit=500&target=" + 
	old_img.Replace(" ", "+");
  bool goto_next_page = true;
  sites_list.Clear();

  while(goto_next_page)
  {
   try {using (WebClient client = new WebClient()) {htmlCode = client.DownloadString(img_usage_list);}}
   catch {Console.WriteLine("No page for this image found!"); goto beginning;}
  
   mc = Regex.Matches(htmlCode, @"\s([-\w]+\.(\w)+\.org)");
   if(mc.Count==0) {Console.WriteLine("No inclusions found!"); goto beginning;}
   foreach (Match m in mc)
   {
	if(!sites_list.Contains(m.Groups[1].Value) && m.Groups[1].Value!="nl.wikipedia.org" && m.Groups[1].Value!="simple.wikipedia.org")
		{sites_list.Add(m.Groups[1].Value);}
   }

   goto_next_page = !htmlCode.Contains("| next 500)");
   if(goto_next_page)
   {
	mc = Regex.Matches(htmlCode, @"\|\s\<a href" + "=\"(/w/index.php" + @"\?title=Special:GlobalUsage&amp;limit=500&amp;from=" + "[^\"]+)\"");
    img_usage_list = "https://" + "commons.wikimedia.org" + mc[0].Groups[1].Value.Replace("&amp;", "&");
   }
  }


  //PROCESSING
  try
  {
   Site wiki_site; PageList inclusions; int site_counter = 1, page_counter; string lang;
   foreach (string st_f in sites_list)
   {
    string st = st_f;
    rejects = 0;
    Console.WriteLine("\n" + site_counter + ". " + st);

    try_connect_again:
    try {wiki_site = new Site("https://" + st, username, password);}
    catch(System.Net.WebException)
    {
     Console.WriteLine("No internet connection or wrong sitename!");
     Console.WriteLine("Check the connection and type sitename again (type 'skip' to skip):");
     st = Console.ReadLine();
     if(st=="skip") {continue;}
     goto try_connect_again;
    }
    catch(DotNetWikiBot.WikiBotException)
    {Console.WriteLine("Failed to login! This site will be skipped."); continue;}

    inclusions = new PageList(wiki_site);
    inclusions.FillFromPagesUsingImage(old_img);
    Console.WriteLine("\t" + inclusions.Count() + " inclusions found" + "\n");
    lang = GetLang(st);
  
    page_counter = 1;
    foreach (Page wiki_page in inclusions)
    {
     Console.WriteLine("\t" + page_counter + ". " + wiki_page.title);
	 
	 if(st_f=="www.wikidata.org")
	 {
	  System.Windows.Forms.Clipboard.SetText(new_img);
	  System.Diagnostics.Process.Start("http://" + "www.wikidata.org/wiki/" + wiki_page.title);
	  continue;
	 }

     Console.Write("\t");
     wiki_page.Load();
     if(wiki_page.Exists() && (wiki_page.text.Contains(old_img) || wiki_page.text.Contains(old_img_underlined)))
     {
      wiki_page.text = wiki_page.text.Replace(old_img, new_img);
      wiki_page.text = wiki_page.text.Replace(old_img_underlined, new_img);
      Console.Write("\t");
      try {edits++; wiki_page.Save(summary[lang], true); if(st.StartsWith("pt.")) {Thread.Sleep(10000);}}
      catch (DotNetWikiBot.WikiBotException)
      {
       edits--;
       Console.WriteLine("\t\t" + "Site did not allow this edit");
       if(++rejects > 3)
       {Console.WriteLine("\t" + "To many rejects. This site will be skipped"); break;}
      }
	  catch(System.Net.WebException e)
	  {
	   Console.WriteLine("\t\t" + "Network error: ");
	   Console.WriteLine("\t\t" + e.Message);
	   Console.WriteLine("\t\t" + "Waiting one minute");
	   Thread.Sleep(60000);
	  }
	  catch(Exception e) {Console.WriteLine("\t\tUnhandled exception: " + e.GetType()); Console.WriteLine("\t\t" + e.Message);}
     }
     else {Console.WriteLine("\t\t" + "Image not present");}

     page_counter++;
    }
    site_counter++;
   }
  }

  catch(Exception e)
  {
   Console.WriteLine("Unhandled exception: " + e.GetType());
   Console.WriteLine(e.Message);
   Console.ReadKey();
  }

  //WRITING TO LOG
  finally
  {
   if(edits>=3)
   {
    if(old_img.Length>64) {old_img = old_img.Substring(0,60) + "...";}
    if(new_img.Length>64) {new_img = new_img.Substring(0,60) + "...";}

    System.IO.StreamWriter outfile = new System.IO.StreamWriter("log.txt", true);
    outfile.Write("\r\n" + old_img);
    int tabs = (old_img.Length)/8;
    tabs = 8 - tabs;
    while (tabs!=0) {outfile.Write("\t"); tabs--;}

    outfile.Write(new_img);
    tabs = (new_img.Length)/8;
    tabs = 8 - tabs;
    while (tabs!=0) {outfile.Write("\t"); tabs--;}

    outfile.Write(DateTime.Now.ToString("dd.MM.yyyy (HH:mm)"));
    outfile.Write("\t{0,4}", edits);
    outfile.Close();
   }
   Console.WriteLine("DONE! - " + edits + " edits done\n");
  }
  if(args.Length>=3) {return;}
  goto beginning;
 }

 
 //CHOSING LANGUAGE OF EDIT SUMMARY
 public static string GetLang(string site)
 {
  string[] rulangs = {"ru.", "kk.", "hy.", "uz.", "ky.", "tg.", "crh.", "gag.", "tt.", "av.", 
                         "kbd.", "ab.", "ba.", "bxr.", "os.", "kv.", "krc.", "mrj.", "lbe.", "lez.", 
                         "mdf.", "ce.", "mhr.", "koi.", "sah.", "cu.", "tyv.", "udm.", "xal.", "cv.", 
			 "myv.", "ady"};
  string[] delangs = {"de.", "lb.", "li.", "als.", "bar.", "ksh."};
  string[] pllangs = {"pl.", "szl.", "csb.", "hsb.", "dsb."};
  string[] srlangs = {"sr.", "sh.", "bs.", "mk."};

  if (site.StartsWith("uk.") || site.StartsWith("rue.")) {return "uk";}
  else if (site.StartsWith("be.") || site.StartsWith("be-tarask.")) {return "be";}
  else if (site.StartsWith("hr.") || site.StartsWith("sl.")) {return "hr";}
  else if (site.StartsWith("pt.") || site.StartsWith("es.") || site.StartsWith("gl.")) {return "pt";}
  else if (site.StartsWith("hu.")) {return "hu";}
  else if (site.StartsWith("ar.")) {return "ar";}
  else if (site.StartsWith("fa.")) {return "fa";}
  else if (site.StartsWith("nv.")) {return "nv";}
  else if (Array.IndexOf(srlangs, site.Substring(0,3))>=0) {return "sr";}
  else if (Array.IndexOf(pllangs, site.Substring(0,3))>=0 || Array.IndexOf(pllangs, site.Substring(0,4))>=0) {return "pl";}
  else if (Array.IndexOf(delangs, site.Substring(0,3))>=0 || Array.IndexOf(delangs, site.Substring(0,4))>=0) {return "de";}
  else if (Array.IndexOf(rulangs, site.Substring(0,3))>=0 || Array.IndexOf(rulangs, site.Substring(0,4))>=0) {return "ru";}
  else {return "en";}
 }
}