private async void Page_Loaded(object sender, RoutedEventArgs e)
{
_classicalMusicContext = new ClassicalMusicContext(new Uri("http://www.harrenstein.com/ClassicalMusic/ClassicalMusic.svc"));
_loadingCancellationTokenSource = new CancellationTokenSource();
var composerId = (int)Application.Current.Properties["SelectedComposer"];
var composer = await _classicalMusicContext.Composers.FirstAsync(c => c.ComposerId == composerId, _loadingCancellationTokenSource.Token);
var compositionId = (int)Application.Current.Properties["SelectedComposition"];
var composition = composer.Compositions.FirstOrDefault(c => c.CompositionId == compositionId);
DataContext = composition;
var compositionArray = new Composition[] { composition };
CompositionDataGrid.ItemsSource = compositionArray;
CompositionDataGrid2.ItemsSource = compositionArray;
CommentDataGrid.ItemsSource = compositionArray;
}
private string CompositionToCompositionDisplayName(Composition composition)
{
if (composition == null)
{
return null;
}
var stringBuilder = new StringBuilder();
stringBuilder.Append(composition.Name);
var catalogNumber = composition.CatalogNumbers.FirstOrDefault();
if (catalogNumber != null)
{
stringBuilder.Append(", ");
stringBuilder.Append(catalogNumber.Catalog.Prefix);
stringBuilder.Append(" ");
stringBuilder.Append(catalogNumber.Value);
}
return stringBuilder.ToString();
}
public async static Task ScrapeCompositionDetailPageAsync(string url, Composition composition, ClassicalMusicContext classicalMusicContext)
{
var compositionWebpage = composition.Links.FirstOrDefault(l => l.Url.Contains("klassika.info"));
if (compositionWebpage == null)
{
compositionWebpage = new Link();
compositionWebpage.Compositions.Add(composition);
compositionWebpage.Url = url;
composition.Links.Add(compositionWebpage);
}
var webClient = new WebClient();
var htmlSource = await webClient.DownloadStringTaskAsync(new Uri(url));
var htmlDocument = new HtmlDocument();
htmlDocument.LoadHtml(htmlSource);
var generalInformationHeader = htmlDocument.DocumentNode
.Descendants("h2")
.FirstOrDefault(n => n.InnerText == "Allgemeine Angaben zum Werk:");
if (generalInformationHeader != null)
{
var generalInformationTable = generalInformationHeader.NextSibling.NextSibling;
var tableRows = generalInformationTable.Descendants("tr");
foreach (var tableRow in tableRows)
{
var tableDatas = tableRow
.Elements("td")
.ToArray();
var headerTableData = tableDatas[0];
var valueTableData = tableDatas[1];
var header = HtmlEntity.DeEntitize(headerTableData.InnerText);
var value = HtmlEntity.DeEntitize(valueTableData.InnerText);
if (header == "Tonart:")
{
var key = classicalMusicContext.Keys.FirstOrDefault(k => k.Name == value);
if (key == null)
{
key = new Key();
key.Name = value;
key.Compositions.Add(composition);
composition.Key = key;
classicalMusicContext.Keys.Add(key);
}
if (composition.Key == null)
{
composition.Key = key;
}
}
else if (header == "Widmung:")
{
if (composition.Dedication == null)
{
composition.Dedication = value;
}
}
else if (header == "Besetzung:")
{
var instrumentation = classicalMusicContext.Instrumentations.FirstOrDefault(i => i.Name == value);
if (instrumentation == null)
{
instrumentation = new Instrumentation();
instrumentation.Name = value;
instrumentation.Compositions.Add(composition);
composition.Instrumentation = instrumentation;
classicalMusicContext.Instrumentations.Add(instrumentation);
}
if (composition.Instrumentation == null)
{
composition.Instrumentation = instrumentation;
}
}
else if (header == "Uraufführung:")
{
if (composition.Premiere == null)
{
composition.Premiere = value;
}
}
else if (header == "Entstehungszeit:")
{
if (composition.Dates == null)
{
composition.Dates = value;
}
}
else if (header == "Anlass:")
{
if (composition.Occasion == null)
{
composition.Occasion = value;
}
}
else if (header == "Bemerkung:")
{
if (composition.Comment == null)
{
composition.Comment = value;
}
}
}
}
var movementsHeader = htmlDocument.DocumentNode
.Descendants("h2")
.FirstOrDefault(n => n.InnerText == "Sätze:");
if (movementsHeader != null)
{
var movementTable = movementsHeader.NextSibling.NextSibling;
var tableRows = movementTable.Descendants("tr");
foreach (var tableRow in tableRows)
{
var movementNumberTableData = tableRow
.Elements("td")
.ElementAt(0);
var movementNameTableData = tableRow
.Elements("td")
.ElementAt(1);
short movementNumber;
if (!short.TryParse(HtmlEntity.DeEntitize(movementNumberTableData.InnerText).Replace(". Satz:", ""), out movementNumber))
{
var tableRowsArray = tableRows.ToArray();
var tableRowIndex = Array.IndexOf(tableRowsArray, tableRow);
movementNumber = Convert.ToInt16(tableRowIndex + 1);
}
var movementName = HtmlEntity.DeEntitize(movementNameTableData.InnerText);
var movement = composition.Movements.FirstOrDefault(m => m.Number == movementNumber);
if (movement == null)
{
movement = new Movement();
movement.Name = movementName;
movement.Composition = composition;
movement.Number = movementNumber;
composition.Movements.Add(movement);
}
}
}
}
public async static Task ScrapeCompositionsPageAsync(string url, Composer composer, ClassicalMusicContext classicalMusicContext, IProgress<double> progress = null, CancellationToken? cancellationToken = null)
{
var webClient = new WebClient();
var htmlSource = await webClient.DownloadStringTaskAsync(new Uri(url));
var htmlDocument = new HtmlDocument();
htmlDocument.LoadHtml(htmlSource);
var tables = htmlDocument.DocumentNode.Descendants("table");
var table = tables.FirstOrDefault(n =>
{
var classAttribute = n.Attributes["class"];
if (classAttribute == null || classAttribute.Value != "wv")
{
return false;
}
return true;
});
if (table == null)
{
return;
}
var tableRows = table.Elements("tr").ToArray();
var increment = 1d;
foreach (var tableRow in tableRows)
{
if (cancellationToken != null && cancellationToken.Value.IsCancellationRequested)
{
return;
}
var tableDatas = tableRow.Elements("td").ToArray();
var nameTableData = tableDatas[1];
var compositionName = HtmlEntity.DeEntitize(nameTableData.InnerText);
var compositionAnchor = nameTableData.Element("a");
var hypertextReference = (HtmlAttribute)null;
var composition = (Composition)null;
if (compositionAnchor != null)
{
hypertextReference = compositionAnchor.Attributes["href"];
if (hypertextReference != null)
{
composition = composer.Compositions.FirstOrDefault(c => c.Links.Any(l => l.Url == $"http://www.klassika.info/{hypertextReference.Value}"));
}
}
var catalogTableData = tableDatas[2];
var catalogString = HtmlEntity.DeEntitize(catalogTableData.InnerText);
var dateTableData = tableDatas[4];
var compositionDate = HtmlEntity.DeEntitize(dateTableData.InnerText);
var genreTableData = tableDatas[6];
var genreName = TranslateGenre(HtmlEntity.DeEntitize(genreTableData.InnerText).Replace("(", "").Replace(")", ""));
if (composition == null)
{
composition = new Composition();
composition.Name = compositionName;
composition.Composers = new DataServiceCollection<Composer> { composer };
composition.Dates = compositionDate == "?" ? null : compositionDate;
composer.Compositions.Add(composition);
}
var catalogNumberString = (string)null;
var compositionCatalogString = (string)null;
if (!string.IsNullOrWhiteSpace(genreName))
{
var genre = classicalMusicContext.Genres.FirstOrDefault(ct => ct.Name == genreName);
if (genre == null)
{
genre = new Genre();
genre.Name = genreName;
genre.Compositions.Add(composition);
classicalMusicContext.Genres.Add(genre);
}
if (composition.Genre == null)
{
composition.Genre = genre;
}
}
if (!string.IsNullOrWhiteSpace(catalogString))
{
var catalogParts = catalogString.Split(new char[] { ' ' }, 2);
if (catalogParts.Length == 2)
{
compositionCatalogString = catalogParts[0];
catalogNumberString = catalogParts[1];
var catalog = composer.Catalogs.FirstOrDefault(cc => cc.Prefix == compositionCatalogString);
if (catalog == null)
{
catalog = new Catalog();
catalog.Prefix = compositionCatalogString;
catalog.Composer = composer;
composer.Catalogs.Add(catalog);
}
var catalogNumber = catalog.CatalogNumbers.FirstOrDefault(cn => cn.Value == catalogNumberString);
if (catalogNumber == null)
{
catalogNumber = new CatalogNumber();
catalogNumber.Catalog = catalog;
catalogNumber.Compositions.Add(composition);
catalogNumber.Value = catalogNumberString;
catalog.CatalogNumbers.Add(catalogNumber);
}
}
}
if (hypertextReference != null)
{
await ScrapeCompositionDetailPageAsync($"http://www.klassika.info/{hypertextReference.Value}", composition, classicalMusicContext);
}
if (progress != null)
{
progress.Report(increment++ / (double)tableRows.Length);
}
#if DEBUG
Debug.WriteLine(composition.Name);
#endif
}
}
