/// <summary>
/// Get the #tab=summary page
/// </summary>
/// <returns>
/// int the ID of movie in database;
/// </returns>
private string GetGeneralDesc()
{
string result = ""; // default ID
bool downloaded = false;
int count = 0;
HtmlDocument doc = new HtmlDocument();
TheNumbersDataContext dc = new TheNumbersDataContext();
Movy movieDB = new Movy();
List <ReleaseDate> reldateDB = new List <ReleaseDate>();
while (!downloaded)
{
try
{
doc.LoadHtml(Helper.GetHTML(url + @"#tab=summary"));
downloaded = true;
}
catch (Exception e)
{
downloaded = false;
count++;
if (count >= 3)
{
Helper.WriteToLog(ProgramStatus.Error, "error loading the movie");
Helper.WriteToError(url, e.StackTrace, year);
}
else
{
Task.Delay(5000);
}
}
}
if (downloaded)
{
try
{
movieDB.ID = dc.GetMovieID();
// get summary div
var summ = doc.DocumentNode.Descendants("div")
.Single(o => o.HasAttributes && o.Attributes.Any(p => p.Name.Equals("id")) &&
o.Attributes["id"].Value.Equals("summary"));
// title
movieDB.Title = doc.DocumentNode.Descendants("h1").Single(tit => tit.HasAttributes && tit.Attributes.Any(attr => attr.Name.ToLowerInvariant().Equals("itemprop")) && tit.Attributes["itemprop"].Value.ToLowerInvariant().Equals("name"))
.InnerText.Trim();
// rating
var rating = summ.Descendants("table").FirstOrDefault();
if (rating.Descendants("tr").Count() > 1)
{
// rating exist
foreach (var a in rating.Descendants("tr").LastOrDefault().Descendants("a"))
{
if (a.InnerText.ToLowerInvariant().Contains("critics"))
{ // critics
string temprating = Regex.Match(a.InnerText, @"\d+").Value;
movieDB.RTCRating = Int32.Parse(temprating);
}
else
{ // audience
string temprating = Regex.Match(a.InnerText, @"\d+").Value;
movieDB.RTARating = Int32.Parse(temprating);
}
}
}
movieDB.year = this.year;
// general desc
var gen = summ.Descendants("table").LastOrDefault();
foreach (var tr in gen.Descendants("tr"))
{
if (tr.InnerText.ToLowerInvariant().Contains("budget"))
{
var tempbudget = tr.InnerText.Split(':').LastOrDefault()
.Trim().Replace("$", string.Empty).Replace(",", string.Empty).Replace(" ", string.Empty);
movieDB.Budget = Int32.Parse(tempbudget);
}
else if (tr.InnerText.ToLowerInvariant().Contains("domestic"))
{
foreach (var rels in tr.Descendants("td").LastOrDefault()
.InnerHtml.Split(new string[] { "<br>" },
StringSplitOptions.RemoveEmptyEntries))
{
var rel = rels.Split(new string[] { "by" },
StringSplitOptions.RemoveEmptyEntries);
var kind = Regex.Match(rel[0], @"\((.*?)\)").Value.Equals("") ? "" : Regex.Match(rel[0], @"\((.*?)\)").Value;
string reldate = "";
DateTime realreldate = new DateTime();
if (kind.Length > 0)
{
reldate = rel[0].Remove(rel[0].IndexOf(',') - 2, 2).Replace(kind, String.Empty).Trim();
}
else
{
reldate = rel[0].Remove(rel[0].IndexOf(',') - 2, 2).Trim();
}
realreldate = DateTime.ParseExact(reldate, "MMMM d, yyyy", CultureInfo.InvariantCulture);
HtmlNode publisher = rel.Count() > 1 ? HtmlNode.CreateNode(rel[1].Trim()) : null;
ReleaseDate reldateDBtemp = new ReleaseDate();
reldateDBtemp.ReleaseDate1 = realreldate;
reldateDBtemp.Remarks = "Domestic";
reldateDBtemp.Company = publisher == null ? "Unknown" : publisher.InnerText;
reldateDBtemp.Desc = kind;
reldateDB.Add(reldateDBtemp);
}
}
else if (tr.InnerText.ToLowerInvariant().IndexOf("video") == 0)
{
foreach (var rels in tr.Descendants("td").LastOrDefault()
.InnerHtml.Split(new string[] { "<br>" },
StringSplitOptions.RemoveEmptyEntries))
{
var rel = rels.Split(new string[] { "by" },
StringSplitOptions.RemoveEmptyEntries);
var kind = Regex.Match(rel[0], @"\((.*?)\)").Value.Equals("") ? "" : Regex.Match(rel[0], @"\((.*?)\)").Value;
string reldate = "";
DateTime realreldate = new DateTime();
if (kind.Length > 0)
{
reldate = rel[0].Remove(rel[0].IndexOf(',') - 2, 2).Replace(kind, String.Empty).Trim();
}
else
{
reldate = rel[0].Remove(rel[0].IndexOf(',') - 2, 2).Trim();
}
realreldate = DateTime.ParseExact(reldate, "MMMM d, yyyy", CultureInfo.InvariantCulture);
var publisher = rel.Count() > 1 ? HtmlNode.CreateNode(rel[1].Trim()) : null;
ReleaseDate reldateDBtemp = new ReleaseDate();
reldateDBtemp.ReleaseDate1 = realreldate;
reldateDBtemp.Remarks = "Video";
reldateDBtemp.Company = publisher == null ? "Unknown" : publisher.InnerText;
reldateDBtemp.Desc = string.IsNullOrEmpty(kind) ? "Unknown" : kind;
reldateDB.Add(reldateDBtemp);
}
}
else if (tr.InnerText.ToLowerInvariant().IndexOf("mpaa") == 0)
{
var mpaarating = tr.Descendants("td").LastOrDefault().Descendants("a").Single()
.InnerText;
//Console.WriteLine(mpaarating);
movieDB.MPAARating = mpaarating;
}
else if (tr.InnerText.ToLowerInvariant().IndexOf("running") == 0)
{
var runningtime = Int32.Parse(Regex.Match(tr.Descendants("td").LastOrDefault()
.InnerText, @"\d+").Value);
movieDB.RunningTime = runningtime;
}
else if (tr.InnerText.ToLowerInvariant().IndexOf("franchise") == 0)
{
var franchise = tr.Descendants("td").LastOrDefault()
.Descendants("a").Single().InnerText;
movieDB.Franchise = franchise;
}
else if (tr.InnerText.ToLowerInvariant().IndexOf("genre") == 0)
{
var genre = tr.Descendants("td").LastOrDefault()
.Descendants("a").Single().InnerText;
//Console.WriteLine(genre);
movieDB.Genre = genre;
}
else if (tr.InnerText.ToLowerInvariant().IndexOf("compan") == 0)
{
var company = "";
foreach (var a in tr.Descendants("td").LastOrDefault()
.Descendants("a"))
{
company += a.InnerText.Trim();
}
movieDB.Company = company;
}
}
try
{
var check = dc.CheckIfExistsTheSame(movieDB.Title, year);
if (check.Equals("false"))
{
dc.ReleaseDates.InsertAllOnSubmit(reldateDB.Select(rd => { rd.MovieID = movieDB.ID; return(rd); }).ToList());
dc.Movies.InsertOnSubmit(movieDB);
dc.SubmitChanges();
result = movieDB.ID;
}
else
{
result = check;
}
}
catch (Exception e)
{
Helper.WriteToLog(ProgramStatus.Error, "error loading the movie");
Helper.WriteToError(url, e.StackTrace, year);
}
}
catch
{
Console.WriteLine("");
}
}
return(result);
}
partial void UpdateReleaseDate(ReleaseDate instance);
partial void DeleteReleaseDate(ReleaseDate instance);
partial void InsertReleaseDate(ReleaseDate instance);
