void ParserIMDB_MoviePage(string pageContent)
{
Movie movie = new Movie();
using (StringReader stringReader = new StringReader(pageContent))
{
List<string> Lines = new List<string>();
string line;
while ((line = stringReader.ReadLine()) != null)
Lines.Add(line.Trim());
for (int i = 0; i < Lines.Count - 1; i++)
{
#region Name
if (movie.Name == null)
{
//string mask = @"<meta property='og:title' content=""*""";
string mask = @"<h1 class=""header""> <span class=""itemprop"" itemprop=""name"">*</span>";
if (Lines[i].EqualByStrMask(mask))
movie.Name = Lines[i].MaskedStrCopy(mask).Trim();
}
#endregion
#region Name2
if (movie.Name2 == null)
{
if (Lines[i] == @"<br/><span class=""title-extra"" itemprop=""name"">")
for (int ii = i + 1; ii < Lines.Count - 1; ii++)
{
string mask = @"""*""";
if (Lines[ii].EqualByStrMask(mask))
{
movie.Name2 = Lines[ii].MaskedStrCopy(mask).Trim();
break;
}
}
if (Lines[i].IndexOf("Also Known As:") > -1)
for (int ii = i + 1; ii < Lines.Count - 1; ii++)
{
string mask = @"<h4 class=""inline"">Also Known As:</h4> ";
if (Lines[ii].EqualByStrMask(mask))
{
movie.Name2 = Lines[ii].MaskedStrCopy(mask).Trim();
break;
}
}
}
#endregion
#region Year#1
if (movie.Year == null)
{
string mask = @"/?ref_=tt_ov_inf"" >";
if (Lines[i].IndexOf(mask) > -1)
movie.Year = Lines[i].Substring(Lines[i].IndexOf(mask) + mask.Length, Lines[i].IndexOf("</a>)</span>") - Lines[i].IndexOf(mask) - mask.Length);
}
#endregion
#region PicUrl&PicBin
if (movie.PicUrl == null)
{
if (Lines[i] == @"<div class=""image"">")
for (int ii = i + 1; ii < Lines.Count - 1; ii++)
{
string mask = @"src=""http://*.jpg""";
if (Lines[ii].EqualByStrMask(mask))
{
movie.PicUrl = "http://" + Lines[ii].MaskedStrCopy(mask).Trim() + ".jpg";
byte[] imageBytes = new WebClient().DownloadData(movie.PicUrl);
movie.PicBin = new MemoryStream(imageBytes);
break;
}
}
}
#endregion
#region Genre
{
if ((Lines[i].IndexOf(@"<a href=""/genre/") > -1)
&& (Lines[i].EndsWith(@"</span></a>")))
{
if (movie.Genre == null)
movie.Genre = "";
string mask = @"""genre"">*</span>";
string genre = Lines[i].MaskedStrCopy(mask).Trim();
if (movie.Genre.IndexOf(genre) == -1)
movie.Genre = (movie.Genre + "," + genre).SpecialTrim(",");
}
}
#endregion
#region Runtime
if (movie.Runtime == TimeSpan.FromMilliseconds(0))
{
if ((Lines[i].StartsWith(@"<time itemprop=""duration"" datetime=""")
&& (Lines[i].EndsWith(@""" >"))))
{
i++;
movie.Runtime = TimeSpan.FromMinutes(double.Parse(Lines[i].Substring(0, Lines[i].IndexOf(" "))));
}
}
#endregion
#region Ratio
if (movie.Ratio == 0)
{
if (Lines[i].IndexOf(@"<span itemprop=""ratingValue"">") > -1)
{
string mask = @"<strong><span itemprop=""ratingValue"">*</span></strong>";
if (Lines[i].EqualByStrMask(mask))
movie.Ratio = (int)(double.Parse(Lines[i].MaskedStrCopy(mask).Trim(), CultureInfo.InvariantCulture) * 10);
}
}
#endregion
#region Director
if (movie.Director == null)
{
if (Lines[i].IndexOf(@"<div class=""txt-block"" itemprop=""director""") > -1)
for (int ii = i + 1; (ii < Lines.Count - 1) && (Lines[ii].IndexOf("div>") == -1); ii++)
if (Lines[ii].IndexOf(@"<span class=""itemprop"" itemprop=""name"">") > -1)
{
if (movie.Director == null)
movie.Director = "";
string mask = @"<span class=""itemprop"" itemprop=""name"">*</span></a>";
string director = Lines[ii].MaskedStrCopy(mask).Trim();
if (movie.Director.IndexOf(director) == -1)
movie.Director = (movie.Director + "," + director).SpecialTrim(",");
}
}
#endregion
#region Writer
if (movie.Writer == null)
{
if (Lines[i].IndexOf(@"<div class=""txt-block"" itemprop=""creator""") > -1)
for (int ii = i + 1; (ii < Lines.Count - 1) && (Lines[ii].IndexOf("div>") == -1); ii++)
if (Lines[ii].IndexOf(@"<span class=""itemprop"" itemprop=""name"">") > -1)
{
if (movie.Writer == null)
movie.Writer = "";
string mask = @"<span class=""itemprop"" itemprop=""name"">*</span></a>";
string writer = Lines[ii].MaskedStrCopy(mask).Trim();
if (movie.Writer.IndexOf(writer) == -1)
movie.Writer = (movie.Writer + "," + writer).SpecialTrim(",");
}
}
#endregion
#region Actor
if (movie.Actor == null)
{
if (Lines[i].IndexOf(@">Cast<") > -1)
for (int ii = i + 1; (ii < Lines.Count - 1) && (Lines[ii].IndexOf("<script>") == -1); ii++)
if (Lines[ii].EndsWith(@"</span>"))
//if (Lines[ii].IndexOf(@"<span class=""itemprop"" itemprop=""name"">") > -1)
{
if (movie.Actor == null)
movie.Actor = "";
string mask = @""" itemprop='url'> <span class=""itemprop"" itemprop=""name"">*</span>";
string Actor = Lines[ii].MaskedStrCopy(mask).Trim();
if (movie.Actor.IndexOf(Actor) == -1)
movie.Actor = (movie.Actor + "," + Actor).SpecialTrim(",");
}
}
#endregion
#region Desc
if (movie.Desc == null)
{
if (Lines[i].IndexOf(@"plotsummary?ref_=tt_stry_pl"" >Plot Summary</a>") > -1)
{
string mask = @"<a href=""*?ref_=tt_stry_pl"" >Plot Summary</a>";
string url = Lines[i].MaskedStrCopy(mask).Trim();
string innerPageContent = new WebClient().DownloadString(@"http://www.imdb.com" + url);
innerPageContent.SaveToFile(@"c:\txt.txt");
using (StringReader innerStringReader = new StringReader(innerPageContent))
{
List<string> innerLines = new List<string>();
string innerLine;
while ((innerLine = innerStringReader.ReadLine()) != null)
innerLines.Add(innerLine.Trim());
for (int ii = 0; ii < innerLines.Count - 1; ii++)
if (innerLines[ii].IndexOf(@"plotpar") > -1)
{
ii++;
movie.Desc = (movie.Desc + "\n" + innerLines[ii]).SpecialTrim("\n");
}
}
}
}
#endregion
#region Url
if (movie.Url == null)
{
if (Lines[i].IndexOf(@"plotsummary?ref_=tt_stry_pl"" >Plot Summary</a>") > -1)
{
string mask = @"<a href=""*plotsummary?ref_=tt_stry_pl"" >Plot Summary</a>";
movie.Url = @"http://www.imdb.com" + Lines[i].MaskedStrCopy(mask).Trim();
}
}
#endregion
}
}
Movies.Add(movie);
}
void ParserIMDB_SearchResultPage(string pageContent)
{
pageContent.SaveToFile(@"c:\txt.txt");
string searchLine;
using (StringReader stringReader = new StringReader(pageContent))
{
while ((searchLine = stringReader.ReadLine().Trim()) != null)
if (searchLine.StartsWith(@"<tr class=""findResult odd"">"))
break;
}
string[] Lines = searchLine.Split(new string[] { @"<tr class=""findResult" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string item in Lines)
{
Movie movie = new Movie();
movie.Name = item.MaskedStrCopy(@">*</a>", item.LastIndexOf(@"fn_tt_tt")).Trim();
movie.Year = item.Substring(item.IndexOf('('), item.LastIndexOf(')') - item.IndexOf('(') + 1);
movie.Url = @"http://www.imdb.com" + item.MaskedStrCopy(@"<a href=""*?ref").Trim();
movie.PicUrl = item.MaskedStrCopy(@"<img src=""*"" />").Trim();
byte[] imageBytes = new WebClient().DownloadData(movie.PicUrl);
movie.PicBin = new MemoryStream(imageBytes);
SearchResult_IMDB.Add(movie);
}
}
