public static Album ParseAlbum(string htmlData)
{
if (htmlData == "")
{
return(null);
}
try
{
HtmlDocument htmlDoc = new HtmlDocument();
htmlDoc.LoadHtml(htmlData);
var infoTitle = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//h1") == null ? null :
htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//h1").First();
/* ExternalAlbumId */
var externalAlbumIdStr = htmlDoc.DocumentNode.SelectNodes("//head//meta[@property='og:url']") == null ? "" :
htmlDoc.DocumentNode.SelectNodes("//head//meta[@property='og:url']").First().GetAttributeValue("content", "");
int externalAlbumId = Int32.Parse(Path.GetFileName(externalAlbumIdStr));
/* Album name */
var albumNameSection = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//h1").First();
string albumName = "";
if (albumNameSection.LastChild.LastChild.LastChild == null)
{
albumName = albumNameSection.LastChild.InnerText.Trim();
}
else
{
albumName = albumNameSection.LastChild.LastChild.InnerText.Trim();
}
/* Artist */
var artistLinks = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//h1//a");
var artists = new List <Artist>();
foreach (var artistStr in artistLinks)
{
if (artistStr.GetAttributeValue("href", "").Contains("artist"))
{
var artistUrl = "https://www.discogs.com" + artistStr.GetAttributeValue("href", "");
var artistData = RequestHelper.GetPageData(artistUrl);
var artist = ParseArtist(artistData);
if (artist != null)
{
artists.Add(artist);
}
}
}
if (artists.Count == 0)
{
artists.Add(new Artist {
Name = "Unknown", ExternalArtistId = -1
});
}
/* Released */
var releasedStr = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//a[contains(@href, 'year')]") == null ? "" :
htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//a[contains(@href, 'year')]").First().InnerText;
int released = ParseYearFromDate(releasedStr);
/* Format */
var profileDivs = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//div");
var formatStr = "";
if (profileDivs != null)
{
for (int i = 0; i < profileDivs.Count - 1; i++)
{
if (profileDivs[i].InnerHtml.Contains("Format:"))
{
formatStr = profileDivs[i + 1].InnerText.Trim();
break;
i++;
}
}
}
/* Genre, Style */
List <Genre> albumGenres = new List <Genre>();
List <Style> albumStyles = new List <Style>();
var profileHrefs = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//a");
foreach (var profileHref in profileHrefs)
{
if (profileHref.GetAttributeValue("href", "").Contains("genre"))
{
albumGenres.Add(new Genre
{
GenreExternalName = Path.GetFileName(profileHref.GetAttributeValue("href", "")),
GenreName = profileHref.InnerText.Trim()
});
}
if (profileHref.GetAttributeValue("href", "").Contains("style"))
{
albumStyles.Add(new Style
{
StyleExternalName = Path.GetFileName(profileHref.GetAttributeValue("href", "")),
StyleName = profileHref.InnerText.Trim()
});
}
}
/* Country */
var country = htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//a[contains(@href, 'country')]") == null ? "" :
htmlDoc.DocumentNode.SelectNodes("//div[@class='profile']//a[contains(@href, 'country')]").First().InnerText.Trim();
/* numVersions */
var versions = htmlDoc.DocumentNode.SelectNodes("//div[contains(@class, 'm_versions')]");
int numVersions = 1;
if (versions != null)
{
var viewAll = htmlDoc.DocumentNode.SelectNodes("//div[contains(@class, 'm_versions')]//h3//a");
if (viewAll == null)
{
var numVersionsStr = htmlDoc.DocumentNode.SelectNodes("//div[contains(@class, 'm_versions')]//tr[contains(@class, 'card r_tr')]");
numVersions = numVersionsStr.Count;
}
else
{
var allVersionsLink = viewAll.First().GetAttributeValue("href", "");
allVersionsLink = "https://www.discogs.com" + allVersionsLink;
var versionsData = RequestHelper.GetPageData(allVersionsLink);
numVersions = ParseNumVersions(versionsData);
}
}
/* Cyrilic or latin */
bool isCyrilic = !Regex.IsMatch(albumName, @"\P{IsCyrillic}");
if (isCyrilic)
{
albumName = CyrilicToLatin(albumName);
}
Album album = new Album
{
Name = albumName,
Country = country,
Format = formatStr,
ExternalAlbumId = externalAlbumId,
NumVersions = numVersions,
Artist = artists,
Released = released,
Tracklist = new List <Song>(),
IsCyrilic = isCyrilic,
Genres = albumGenres,
Styles = albumStyles
};
/* Tracklist */
var tracklistSections = htmlDoc.DocumentNode.SelectNodes("//div[@class='section tracklist']//table[@class='playlist']//tr[contains(@class, ' tracklist_track track')]");
foreach (var tracklistSection in tracklistSections)
{
var tracklistLink = tracklistSection.SelectNodes("td[@class='track tracklist_track_title ']//a") == null ? "" :
tracklistSection.SelectNodes("td[@class='track tracklist_track_title ']//a").First().GetAttributeValue("href", "");
/*if (tracklistLink == "")
* {
* continue;
* string songName = tracklistSection.SelectNodes("td[@class='track tracklist_track_title ']//a") == null ? "" :
* tracklistSection.SelectNodes("td[@class='track tracklist_track_title ']//a").First().InnerText.Trim();
* Song s = new Song
* {
* Name = songName,
* Genres = "",
* Styles = new List<string>()
* };
* album.Tracklist.Add(s);
* }*/
if (tracklistLink != "")
{
tracklistLink = string.Format("https://www.discogs.com{0}", tracklistLink);
var tracklistData = RequestHelper.GetPageData(tracklistLink);
Song song = ParseSong(tracklistData);
if (song != null)
{
album.Tracklist.Add(song);
}
}
}
return(album);
}
catch (Exception ex)
{
Console.WriteLine("Parse album: " + ex.Message);
return(null);
}
}
