public void ParseTVPage(string fileContent, WebBrowser browser)
{
List <string> tvContent = new List <string>();
int pos;
while (true)
{
pos = fileContent.IndexOf("<!--");
if (pos == -1)
{
break;
}
int pos2 = fileContent.IndexOf("-->", pos);
if (pos2 == -1)
{
break;
}
fileContent = fileContent.Remove(pos, pos2 - pos + 3);
}
string searchValue = "data-asin=\"";
pos = fileContent.IndexOf(searchValue, 0);
string tempContent = fileContent.Substring(pos);
while (true)
{
int pos2 = tempContent.IndexOf(searchValue, 0);
int pos3 = tempContent.IndexOf("</li>", pos2 + 1);
string tvseries = tempContent.Substring(pos2, pos3 - pos2 + 5);
tvContent.Add(tvseries);
tempContent = tempContent.Substring(pos3 + 5);
if (tempContent.IndexOf(searchValue) == -1)
{
break;
}
}
foundItems += tvContent.Count;
tbFound.Text = foundItems.ToString();
foreach (string _tvshow in tvContent)
{
pos = _tvshow.IndexOf("data-asin=\"");
int pos2 = _tvshow.IndexOf("\"", pos + 12);
string dataSin = _tvshow.Substring(pos + 11);
pos2 = dataSin.IndexOf("\"", 0);
dataSin = dataSin.Substring(0, pos2);
Tvseries tvSeries = new Tvseries();
tvSeries.DataSin = dataSin;
pos = _tvshow.IndexOf("out of 5 stars");
if (pos != -1)
{
pos2 = _tvshow.IndexOf("<", pos);
while (pos > 0 && _tvshow[pos] != '>')
{
pos--;
}
if (_tvshow[pos] == '>')
{
tvSeries.Stars = _tvshow.Substring(pos + 1, pos2 - pos - 1);
}
}
pos = _tvshow.IndexOf("src=\"");
pos2 = _tvshow.IndexOf("\"", pos + 6);
tvSeries.Image = _tvshow.Substring(pos + 5);
pos2 = tvSeries.Image.IndexOf("\"", 0);
tvSeries.Image = tvSeries.Image.Substring(0, pos2);
pos = _tvshow.IndexOf("href=\"");
pos2 = _tvshow.IndexOf("\"", pos + 7);
tvSeries.Url = _tvshow.Substring(pos + 6);
pos2 = tvSeries.Url.IndexOf("\"", 0);
tvSeries.Url = tvSeries.Url.Substring(0, pos2);
pos = _tvshow.IndexOf("title=\"");
pos2 = _tvshow.IndexOf("\"", pos + 8);
tvSeries.Title = _tvshow.Substring(pos + 7);
pos2 = tvSeries.Title.IndexOf("\"", 0);
tvSeries.SearchTitle = tvSeries.Title = HttpUtility.HtmlDecode(tvSeries.Title.Substring(0, pos2));
if (tvSeries.Title.ToUpper().IndexOf("THE ") == 0)
{
tvSeries.SearchTitle = tvSeries.Title.Substring(4) + ", " + tvSeries.Title.Substring(0, 3);
}
pos = _tvshow.IndexOf("a-icon-alt\">");
if (pos != -1)
{
pos2 = _tvshow.IndexOf("\"", pos + 11);
tvSeries.Stars = _tvshow.Substring(pos + 12);
pos2 = tvSeries.Stars.IndexOf("<", 0);
tvSeries.Stars = tvSeries.Stars.Substring(0, pos2);
}
pos = 0;
while (pos > -1)
{
pos = _tvshow.IndexOf("class=\"a-size-small a-color-secondary\">", pos + 1);
if (pos != -1)
{
pos2 = _tvshow.IndexOf("<", pos + 1);
string value = _tvshow.Substring(pos + 39, pos2 - pos - 39);
if (tvSeries.Released == 0)
{
int year = -1;
if (int.TryParse(value, out year))
{
tvSeries.Released = year;
}
}
else
{
switch (value)
{
case "CC":
tvSeries.ClosedCaptioned = true;
break;
}
}
}
}
object existingRecord = context.Tvseries.Where(t => t.Released == tvSeries.Released && t.Title == tvSeries.Title).FirstOrDefault <Tvseries>();
if (existingRecord != null)
{
try
{
context.Remove(tvSeries);
}
catch (Exception)
{
}
tvSeries = null;
continue;
}
context.Add(tvSeries);
newItems += 1;
tbNew.Text = newItems.ToString();
context.SaveChanges();
}
var nextButton = browser.Document.GetElementById("pagnNextLink");
if (nextButton == null)
{
MessageBox.Show("Update complete");
rbMovies.Enabled = true;
rbTV.Enabled = true;
btnSearch.Enabled = true;
}
else
{
nextButton.InvokeMember("click");
while (webBrowser1.ReadyState != WebBrowserReadyState.Complete)
{
Application.DoEvents();
}
timer1.Start();
}
}
public void ParseTVDetails(string result)
{
try
{
List <string> genres = new List <string>();
Tvseries tv = tvForDetails[itemsProcessed];
context.Tvseries.Update(tv);
int pos = result.IndexOf("Watch for 0.00 with Prime");
tv.IsPrime = pos != -1;
pos = result.IndexOf("av-icon--amazon_rating");
int pos2;
int pos3;
if (pos != -1)
{
pos = result.IndexOf("(", pos);
pos2 = result.IndexOf(")", pos);
tv.Ratings = result.Substring(pos + 1, pos2 - pos - 1);
}
pos = result.IndexOf("\"imdb-rating-badge\"");
if (pos != -1)
{
pos = result.IndexOf(">", pos);
pos2 = result.IndexOf("<", pos);
try
{
tv.Imdbrating = Convert.ToSingle(result.Substring(pos + 1, pos2 - pos - 1));
}
catch (Exception)
{
}
}
pos = result.IndexOf("\"synopsis\"");
if (pos != -1)
{
pos = result.IndexOf("<p", pos);
if (pos > -1)
{
pos = result.IndexOf(">", pos);
pos2 = result.IndexOf("</p>", pos);
try
{
tv.Plot = result.Substring(pos + 1, pos2 - pos - 1).Replace("Now included with Prime.", "").Trim();
}
catch (Exception)
{
}
}
}
int posProductDetails = result.IndexOf("Product details");
if (posProductDetails != -1)
{
int posOtherFormats = result.IndexOf("Customer Reviews", posProductDetails);
int posTR = -1;
pos = result.IndexOf("Genres", posProductDetails);
if (pos > posOtherFormats)
{
pos = -1;
}
if (pos != -1)
{
posTR = result.IndexOf("<tr", pos);
pos3 = result.IndexOf("</td>", pos);
pos = result.IndexOf("<a", pos);
tv.Genres = string.Empty;
while (pos < pos3 && pos < posTR && pos != -1)
{
if (pos >= posOtherFormats)
{
break;
}
pos = result.IndexOf(">", pos);
pos2 = result.IndexOf("</a>", pos);
try
{
string genre = HttpUtility.HtmlDecode(result.Substring(pos + 1, pos2 - pos - 1));
genres.Add(genre);
genre += ", ";
tv.Genres += genre;
pos = result.IndexOf("<a", pos);
}
catch (Exception)
{
}
}
}
pos = result.IndexOf("Director", posProductDetails);
if (pos != -1)
{
posTR = result.IndexOf("<tr", pos);
pos3 = result.IndexOf("</td>", pos);
pos = result.IndexOf("<a", pos);
if (pos < pos3 && pos < posTR && pos != -1)
{
pos = result.IndexOf(">", pos);
pos2 = result.IndexOf("</a>", pos);
try
{
tv.Director = HttpUtility.HtmlDecode(result.Substring(pos + 1, pos2 - pos - 1));
}
catch (Exception)
{
}
}
}
pos = result.IndexOf("Starring", posProductDetails);
if (pos != -1)
{
posTR = result.IndexOf("<tr", pos);
pos3 = result.IndexOf("</td>", pos);
pos = result.IndexOf("<a", pos);
tv.Starring = string.Empty;
while (pos < pos3 && pos < posTR && pos != -1)
{
if (pos >= posOtherFormats)
{
break;
}
pos = result.IndexOf(">", pos);
pos2 = result.IndexOf("</a>", pos);
try
{
string data = HttpUtility.HtmlDecode(result.Substring(pos + 1, pos2 - pos - 1)) + ", ";
tv.Starring += data;
pos = result.IndexOf("<a", pos);
}
catch (Exception)
{
}
}
if (tv.Starring != null && tv.Starring.Length > 0)
{
tv.Starring = tv.Starring.Replace(" ", " ");
tv.Starring = tv.Starring.Substring(0, tv.Starring.Length - 2);
}
}
pos = result.IndexOf("Supporting actors", posProductDetails);
if (pos != -1)
{
posTR = result.IndexOf("<tr", pos);
pos3 = result.IndexOf("</td>", pos);
pos = result.IndexOf("<a", pos);
tv.SupportingActors = string.Empty;
while (pos < pos3 && pos < posTR && pos != -1)
{
if (pos >= posOtherFormats)
{
break;
}
pos = result.IndexOf(">", pos);
pos2 = result.IndexOf("</a>", pos);
try
{
string data = HttpUtility.HtmlDecode(result.Substring(pos + 1, pos2 - pos - 1)) + ", ";
tv.SupportingActors += data;
pos = result.IndexOf("<a", pos);
}
catch (Exception)
{
}
}
if (tv.SupportingActors != null && tv.SupportingActors.Length > 0)
{
tv.SupportingActors = tv.SupportingActors.Replace(" ", " ");
tv.SupportingActors = tv.SupportingActors.Substring(0, tv.SupportingActors.Length - 2);
}
}
}
context.SaveChanges();
foreach (string _genre in genres)
{
Genre genre = context.Genre.Where(g => g.Name == _genre).SingleOrDefault();
if (genre == null)
{
genre = new Genre();
genre.Name = _genre;
context.Add(genre);
context.SaveChanges();
}
TvseriesGenre tvseriesGenre = new TvseriesGenre();
tvseriesGenre.Genre = genre;
tvseriesGenre.Tvseries = tv;
context.Add(tvseriesGenre);
context.SaveChanges();
}
itemsProcessed++;
lblDetailProgress.Text = $"{itemsProcessed} of {itemsToProcess} Bad URLs: {badURLs}";
if (itemsProcessed < tvForDetails.Count())
{
nextURL = tvForDetails[itemsProcessed].Url;
webBrowser1.Navigate(nextURL);
while (webBrowser1.ReadyState != WebBrowserReadyState.Complete)
{
Application.DoEvents();
}
timer1.Start();
}
else
{
MessageBox.Show("Update complete");
}
}
catch (Exception ex)
{
var error = ex;
}
}