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 ScrapeComposerDetailPageAsync(string url, Composer composer, ClassicalMusicContext classicalMusicContext, IProgress<double> progress = null, CancellationToken? cancellationToken = null)
{
var composerLink = composer.Links.FirstOrDefault(l => l.Url.Contains("klassika.info"));
if (composerLink == null)
{
composerLink = new Link();
composerLink.Composers.Add(composer);
composerLink.Url = url;
composer.Links.Add(composerLink);
}
var urlParts = url.Split('/');
if (urlParts.Length < 2)
{
return;
}
var composerKey = urlParts.ElementAt(urlParts.Length - 2);
await ScrapeCompositionsPageAsync($"http://www.klassika.info/Komponisten/{composerKey}/wv_abc.html", composer, classicalMusicContext, progress, cancellationToken);
}