/// <summary>
/// Main Parse method of the Photo Gallery Page
/// </summary>
/// <param name="movie">Movie instance that is populated</param>
/// <param name="documentNode">Document Node of the photo gallery page</param>
/// <param name="settings">Object containing Data Fetch settings</param>
public static void Parse(Movie movie, HtmlNode documentNode, ProductionDataFetchSettings settings)
{
if (documentNode != null)
{
HtmlNode mediaIndexNode = documentNode.QuerySelector("#media_index_content");
movie.MediaImages = new List <Image>();
if (mediaIndexNode != null)
{
HtmlNode[] allImageNodes = mediaIndexNode.QuerySelectorAll("img").ToArray();
if (allImageNodes != null && allImageNodes.Length != 0)
{
int endIndex = allImageNodes.Length;
if (settings.MediaImagesFetchCount < endIndex)
{
endIndex = settings.MediaImagesFetchCount;
}
for (int i = 0; i < endIndex; i++)
{
HtmlNode imageNode = allImageNodes[i];
Image image = new Image
{
Title = imageNode.GetAttributeValue("title", string.Empty),
URL = IMDBImageHelper.NormalizeImageUrl(imageNode.GetAttributeValue("src", string.Empty))
};
if (settings.FetchImageContents)
{
image.Content = IMDBImageHelper.GetImageContent(image.URL);
}
movie.MediaImages.Add(image);
}
}
}
}
}
private static void CleanupImageContents(ProductionDataFetchSettings settings, Production production)
{
if (!settings.FetchImageContents && production != null)
{
if (production.Poster != null)
{
production.Poster.Content = null;
}
if (production.MediaImages != null)
{
foreach (Image image in production.MediaImages)
{
image.Content = null;
}
}
if (production is Movie)
{
Movie movie = (Movie)production;
if (movie.Credits != null)
{
foreach (Credit credit in movie.Credits)
{
RemoveImageContentsFromPerson(credit.Person);
}
}
}
}
}
/// <summary>
/// Gets Production Information iresspective of type
/// </summary>
/// <param name="id">ID of the production</param>
/// <param name="settings">Object containing Data Fetch settings</param>
/// <returns>Production instance containing retreived information</returns>
public Production GetProduction(long id, ProductionDataFetchSettings settings)
{
GetProductionDetailsRequest request = new GetProductionDetailsRequest {
ID = id, Settings = settings
};
GetProductionDetailsResponse response = jmAppClientProvider.CallAction <GetProductionDetailsResponse>(ActionNameConstants.GetProductionDetails, request);
return(response.Production);
}
/// <summary>
/// Gets Production Information iresspective of type
/// </summary>
/// <param name="id">ID of the production</param>
/// <param name="settings">Object containing Data Fetch settings</param>
/// <returns>Production instance containing retreived information</returns>
public Production GetProduction(long id, ProductionDataFetchSettings settings)
{
using (JMoviesEntities entities = new JMoviesEntities())
{
Production production = entities.Production.FirstOrDefault(e => e.ID == id);
if (settings.MediaImagesFetchCount != 0 && production != null)
{
//fetch poster
production.Poster = entities.Image.FirstOrDefault(e => e.ID == production.PosterID);
//fetch images
int imageCount = settings.MediaImagesFetchCount;
if (imageCount <= 0)
{
//default value
imageCount = 5;
}
production.MediaImages = entities.Image.Where(e => e.ProductionID == production.ID).Take(imageCount).ToArray();
}
if (production is Movie)
{
Movie movie = production as Movie;
int castCount = settings.CastFetchCount;
if (castCount <= 0)
{
//default value
castCount = 5;
}
movie.Credits = entities.Credit.Include(e => (e as ActingCredit).Characters)
.Where(e => e.ProductionID == production.ID).Take(castCount).ToArray();
if (movie.Credits != null)
{
foreach (Credit credit in movie.Credits)
{
credit.Person = entities.Person.FirstOrDefault(e => e.ID == credit.PersonID);
credit.Person.PrimaryImage = entities.Image.FirstOrDefault(e => e.ID == credit.Person.PrimaryImageID);
}
}
movie.Genres = entities.Genre.Where(e => e.ProductionID == production.ID).ToArray();
}
CleanupImageContents(settings, production);
return(production);
}
}
/// <summary>
/// Gets Production Information iresspective of type
/// </summary>
/// <param name="id">ID of the production</param>
/// <param name="settings">Object containing Data Fetch settings</param>
/// <returns>Production instance containing retreived information</returns>
public Production GetProduction(long id, ProductionDataFetchSettings settings)
{
if (id == default(long))
{
throw new JMException("IMDbIDEmpty");
}
else if (settings == null)
{
throw new JMException("SettingsEmpty");
}
Movie movie = new Movie();
string url = IMDbConstants.BaseURL + IMDbConstants.MoviesPath + IMDbConstants.MovieIDPrefix + id.ToString().PadLeft(IMDbConstants.IMDbIDLength, '0');
HtmlDocument htmlDocument = HtmlHelper.GetNewHtmlDocument();
WebRequest webRequest = HttpHelper.InitializeWebRequest(url);
using (Stream stream = HttpHelper.GetResponseStream(webRequest))
{
htmlDocument.Load(stream, Encoding.UTF8);
}
HtmlNode documentNode = htmlDocument.DocumentNode;
//Parse and verify IMDb ID Meta Tag
HtmlNode idMetaTag = documentNode.QuerySelector("meta[property='pageId']");
if (idMetaTag != null)
{
movie.IMDbID = Regex.Replace(idMetaTag.Attributes["content"].Value, IMDbConstants.MovieIDPrefix, string.Empty).ToLong();
}
else
{
return(null);
}
MoviePageDetailsHelper.Parse(this, ref movie, documentNode, url, settings);
return(movie);
}
/// <summary>
/// Gets TV Series information
/// </summary>
/// <param name="id">ID of the TV Series</param>
/// <param name="settings">Object containing Data Fetch settings</param>
/// <returns>TV Series instance containing retreived information</returns>
public TVSeries GetTvSeries(long id, ProductionDataFetchSettings settings)
{
return(GetProduction(id, settings) as TVSeries);
}
/// <summary>
/// Gets Movie information
/// </summary>
/// <param name="id">ID of the movie</param>
/// <param name="settings">Object containing Data Fetch settings</param>
/// <returns>Movie instance containing retreived information</returns>
public Movie GetMovie(long id, ProductionDataFetchSettings settings)
{
return(GetProduction(id, settings) as Movie);
}
/// <summary>
/// Main Parse method of the Movie Page
/// </summary>
/// <param name="providerInstance">Instance reference of the IMDbScraperDataProvider</param>
/// <param name="movie">Movie instance that is populated</param>
/// <param name="documentNode">Document Node of the movie page</param>
/// <param name="moviePageUrl">URL of the movie page</param>
/// <param name="settings">Object containing Data Fetch settings</param>
/// <returns>If scraping was successful or not</returns>
public static bool Parse(IMDbScraperDataProvider providerInstance, ref Movie movie, HtmlNode documentNode, string moviePageUrl, ProductionDataFetchSettings settings)
{
HtmlNode titleTypeTag = documentNode.QuerySelector("meta[property='og:type']");
if (titleTypeTag != null && titleTypeTag.Attributes["content"].Value == IMDbConstants.TVSeriesOgType)
{
//Initialize movie as TV Series
movie = new TVSeries
{
IMDbID = movie.IMDbID
};
}
//Parse Title
HtmlNode titleWrapper = documentNode.QuerySelector(".title_wrapper");
if (titleWrapper != null)
{
movie.Title = titleWrapper.QuerySelector("h1").InnerText.Prepare();
if (IMDbConstants.MovieYearRegex.IsMatch(movie.Title))
{
Match yearMatch = IMDbConstants.MovieYearRegex.Match(movie.Title);
movie.Year = yearMatch.Groups[2].Value.Trim().ToInteger();
movie.Title = yearMatch.Groups[1].Value.Trim();
}
HtmlNode originalTitleNode = titleWrapper.QuerySelector(".originalTitle");
if (originalTitleNode != null)
{
movie.OriginalTitle = originalTitleNode.InnerText.Prepare();
}
foreach (HtmlNode titleLink in titleWrapper.QuerySelectorAll("a"))
{
if (titleLink.OuterHtml.Contains("/releaseinfo"))
{
Match yearMatch = IMDbConstants.MovieYearRegex.Match(titleLink.InnerText.Prepare());
if (yearMatch.Success)
{
movie.Year = yearMatch.Groups[2].Value.Trim().ToInteger();
if (yearMatch.Groups.Count > 3)
{
string endYearString = yearMatch.Groups[3].Value.Trim();
if (!string.IsNullOrEmpty(endYearString))
{
(movie as TVSeries).EndYear = yearMatch.Groups[3].Value.Trim().ToInteger();
}
}
}
}
}
}
else
{
return(false);
}
HtmlNode posterNode = documentNode.QuerySelector(".poster img");
if (posterNode != null)
{
movie.Poster = new Image
{
Title = posterNode.GetAttributeValue("title", string.Empty),
URL = IMDBImageHelper.NormalizeImageUrl(posterNode.GetAttributeValue("src", string.Empty))
};
if (settings.FetchImageContents)
{
movie.Poster.Content = IMDBImageHelper.GetImageContent(movie.Poster.URL);
}
}
//Parse Summary
HtmlNode summaryWrapper = documentNode.QuerySelector(".plot_summary_wrapper");
List <Credit> credits = new List <Credit>();
if (summaryWrapper != null)
{
HtmlNode summaryText = summaryWrapper.QuerySelector(".summary_text");
if (summaryText != null)
{
movie.PlotSummary = summaryText.FirstChild.InnerText.Prepare();
if (movie.PlotSummary.StartsWith(IMDbConstants.EmptyPlotText))
{
movie.PlotSummary = string.Empty;
}
}
foreach (HtmlNode creditSummaryNode in summaryWrapper.QuerySelectorAll(".credit_summary_item"))
{
List <Credit> summaryCredits = SummaryCastHelper.GetCreditInfo(creditSummaryNode);
if (summaryCredits != null && summaryCredits.Count > 0)
{
credits.AddRange(summaryCredits);
}
}
}
else
{
return(false);
}
//Parse Story Line
HtmlNode storyLineSection = documentNode.QuerySelector("#titleStoryLine");
if (storyLineSection != null)
{
SummaryStorylineHelper.Parse(movie, storyLineSection);
}
//Parse Details Section
HtmlNode detailsSection = documentNode.QuerySelector("#titleDetails");
if (detailsSection != null)
{
MoviePageDetailsHelper.ParseDetailsSection(movie, detailsSection);
}
if (!settings.FetchDetailedCast)
{
//Parse Cast Table
HtmlNode castListNode = documentNode.QuerySelector(".cast_list");
ParseCastList(movie, credits, castListNode);
}
else
{
//Fetch credits through full credits page
string fullCreditsUrl = moviePageUrl + "/" + IMDbConstants.FullCreditsPath;
WebRequest fullCreditsPageRequest = HttpHelper.InitializeWebRequest(fullCreditsUrl);
HtmlDocument creditsPageDocument = HtmlHelper.GetNewHtmlDocument();
using (Stream stream = HttpHelper.GetResponseStream(fullCreditsPageRequest))
{
creditsPageDocument.Load(stream, Encoding.UTF8);
}
HtmlNode fullCreditsPageDocumentNode = creditsPageDocument.DocumentNode;
HtmlNode fullCreditsPageCastListNode = fullCreditsPageDocumentNode.QuerySelector(".cast_list");
ParseCastList(movie, credits, fullCreditsPageCastListNode);
movie.Credits = credits;
}
#region Parse Relase Info Page
string releaseInfoURL = moviePageUrl + "/" + IMDbConstants.ReleaseInfoPath;
WebRequest releaseInfoPageRequest = HttpHelper.InitializeWebRequest(releaseInfoURL);
HtmlDocument releaseInfoPageDocument = HtmlHelper.GetNewHtmlDocument();
using (Stream stream = HttpHelper.GetResponseStream(releaseInfoPageRequest))
{
releaseInfoPageDocument.Load(stream, Encoding.UTF8);
}
ReleaseInfoPageHelper.Parse(movie, releaseInfoPageDocument);
#endregion
#region Parse Ratings
HtmlNode ratingsWrapper = documentNode.QuerySelector(".imdbRating");
if (ratingsWrapper != null)
{
HtmlNode ratingNode = ratingsWrapper.QuerySelector("span[itemprop='ratingValue']");
HtmlNode ratingCountNode = ratingsWrapper.QuerySelector("span[itemprop='ratingCount']");
movie.Rating = new Rating(DataSourceTypeEnum.IMDb, movie);
movie.Rating.Value = double.Parse(ratingNode.InnerText.Prepare().Replace('.', ','));
movie.Rating.RateCount = ratingCountNode.InnerText.Prepare().Replace(",", string.Empty).ToLong();
}
#endregion
#region Parse Photo Gallery Page
if (settings.MediaImagesFetchCount > 0)
{
string photoGalleryURL = moviePageUrl + "/" + IMDbConstants.PhotoGalleryPath;
WebRequest photoGalleryPageRequest = HttpHelper.InitializeWebRequest(photoGalleryURL);
HtmlDocument photoGalleryPageDocument = HtmlHelper.GetNewHtmlDocument();
using (Stream stream = HttpHelper.GetResponseStream(photoGalleryPageRequest))
{
photoGalleryPageDocument.Load(stream, Encoding.UTF8);
}
PhotoGalleryPageHelper.Parse(movie, photoGalleryPageDocument?.DocumentNode, settings);
}
#endregion
return(true);
}
