public void DumpPage(int year, int page)
{
var url = string.Format("http://www.mymovies.it/film/{0}/?pagina={1}", year, page);
Console.WriteLine(url);
// var pageHtml = SimpleWebRequest.GetHtmlRobust(url, Encoding: "iso-8859-1");
web.Headers[HttpRequestHeader.UserAgent] = SimpleWebRequest.DefaultUserAgent;
var pageHtml = web.DownloadString(url);
if (pageHtml.Contains("Si è verificato un errore nella pagina, riprova a collegarti facendo clic all'indirizzo qui sotto"))
{
throw new EndOfPagesException();
}
var parser = new QuickTextParser(pageHtml.BetweenS("<table class=\"struttura\" border=\"0\" cellpadding=\"0\" cellspacing=\"0\">", "<!--***** Inizio Sinistra *****-->"), "<a", "<div style=\"clear:both; height:10px;\"></div>");
foreach (var movieHtmlVs in parser)
{
var movieHtml = "<a" + movieHtmlVs.AsString;
var node = movieHtml.AsHtmlNode();
var movie = new Movie()
{
MyMoviesId = int.Parse(node.FindAll("script").ElementAt(1).InnerText.TryCapture(@"schiarisci(\d+)\(")),
Title = (node.FindSingle("h2 a") ?? node.FindSingle("a")).GetText(),
Rating = TryGet(() => Single.Parse(movieHtml.BetweenS("valutazione media tra critica e pubblico: ", " stelle").Replace(',', '.'), System.Globalization.NumberFormatInfo.InvariantInfo)),
Genre = TryGet(() => movieHtml.TryCapture(@"Genere\s*[^>]*>([^<]+)")),
Year = year,
Summary = TryGet(() => node.FindSingle("p[style='margin-top:5px']").GetText()),
ImageCode = TryGet(() => movieHtml.TryCapture(@"http\://pad\.mymovies\.it/filmclub\/([\d/]+)\/imm2")),
TrailerCode = TryGet(() => movieHtml.TryCapture(@"videotrailer_centrale\.asp\?codicefilm=([\d/]+)&")),
ShortDescription = TryGet(() => node.FindSingle("h3").GetText()),
ShortName = TryGet(() => node.FindSingle("h2 a").GetAttributeValue("href", string.Empty).TryCapture(@"film/\d{4}/(.+?)/"))
};
if (db.Movies.Any(x => x.MyMoviesId == movie.MyMoviesId))
{
Console.WriteLine("Duplicato: {0} {1}", movie.MyMoviesId, movie.Title);
continue;
}
Console.WriteLine(movie.Title);
var suggested = movieHtml.TryCapture(">Consigliato: (.+?)<");
switch (suggested)
{
case "Assolutamente No": movie.Suggestion = 1; break;
case "No": movie.Suggestion = 2; break;
case "Nì": movie.Suggestion = 3; break;
case "Sì": movie.Suggestion = 4; break;
case "Assolutamente Sì": movie.Suggestion = 5; break;
case null: break;
default: throw new Exception();
}
var actorsHtmls = new QuickTextParser(node.FindSingle(".linkblu").InnerHtml, "http://www.mymovies.it/biografia/?a=", "</a>");
foreach (var act in actorsHtmls)
{
var s = act.AsString;
var actorCode = int.Parse(s.TryCapture(@"(\d+)"));
var actorName = s.TryCapture(@">([^<]+)").DeEntitize();
var actor = db.Actors.Where(x => x.MyMoviesId == actorCode).SingleOrDefault();
if (actor == null)
{
actor = new Actor();
actor.MyMoviesId = actorCode;
actor.Name = actorName;
db.Actors.AddObject(actor);
db.SaveChanges();
}
movie.Actors.Add(actor);
}
var countries = movieHtml.TryCapture(@"produzione ([^<]+),\s*<a");
if (countries != null)
{
foreach (var cntr in countries.DeEntitize().Split(','))
{
var name = cntr.Trim();
var country = db.Countries.Where(x => x.Name == name).SingleOrDefault();
if (country == null)
{
country = new Country();
country.Name = name;
db.Countries.AddObject(country);
}
movie.Countries.Add(country);
db.SaveChanges();
}
}
var directorIdString = movieHtml.TryCapture(@"biografia/\?r=(\d+)");
if (directorIdString != null)
{
var directorId = int.Parse(directorIdString);
var directorName = movieHtml.TryCapture(@"biografia/\?r=\d+\x22>(.+?)<").DeEntitize();
var dir = db.Directors.Where(x => x.MyMoviesId == directorId).SingleOrDefault();
if (dir == null)
{
dir = new Director();
dir.MyMoviesId = directorId;
dir.Name = directorName;
db.Directors.AddObject(dir);
}
movie.Director = dir;
}
if (movie.EntityState == System.Data.EntityState.Detached)
{
db.Movies.AddObject(movie);
}
// db.Movies.AddObject(movie);
db.SaveChanges();
}
}
public IEnumerable <TvProgram> GetProgramsFromSavedHtmlIfAvailable()
{
if (_html == null)
{
return(Enumerable.Empty <TvProgram>());
}
var p = new QuickTextParser(_html.BetweenS("<body", "class=\"piedipagina\""),
"<div style=\"font-size:180%; text-align:right; ");
var programs = new List <TvProgram>();
bool midnightPassed = false;
double time_prev = 0;
double time_curr = 0;
TvProgram previousProgram = null;
foreach (var t in p)
{
if (!t.Contains("<strong>Ore"))
{
continue;
}
TvProgram prog;
try
{
prog = TvProgram.FromHtml(t, Day, Channel);
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine("Errore TvProgram: " + ex.Message);
continue;
}
time_curr = prog.TimeOfDay.TotalSeconds;
if (time_curr < time_prev)
{
if (midnightPassed)
{
throw new InvalidDataException("Channels are not correctly ordered.");
}
else
{
midnightPassed = true;
}
}
if (midnightPassed)
{
prog.FixToNextDay();
}
time_prev = time_curr;
if (previousProgram != null)
{
previousProgram.Duration = prog.Date - previousProgram.Date;
}
previousProgram = prog;
// if (!midnightPassed && prog.time)
programs.Add(prog);
//if (t.Contains("ATTENZIONE: i seguenti programmi si riferiscono al giorno dopo.")) ;
}
return(programs);
}