/// <summary>
/// Parse the Detail Page returned from the Allmusic Scraper
/// </summary>
/// <param name="strUrl">URL of artist details page</param>
/// <returns>True is scrape was sucessful</returns>
public bool Parse(string strUrl)
{
var mainPage = new HtmlWeb().Load(strUrl);
// moods
var moods = string.Empty;
var moodNodes = mainPage.DocumentNode.SelectNodes(@"//section[@class=""moods""]/ul/*");
if (moodNodes != null)
{
moods = moodNodes.Aggregate(moods, (current, mood) => current + (AllmusicSiteScraper.CleanInnerText(mood) + ", "));
moods = moods.TrimEnd(new[] { ',', ' ' });
}
// artist name
var artistName = AllmusicSiteScraper.CleanInnerText(mainPage.DocumentNode.SelectSingleNode(@"//h2[@clas=""artist-name""]"));
// artist image URL
var artistImg = AllmusicSiteScraper.CleanAttribute(mainPage.DocumentNode.SelectSingleNode(@"//div[@class=""artist-image""]/img"), "src");
//years active
var yearsActive = AllmusicSiteScraper.CleanInnerText(mainPage.DocumentNode.SelectSingleNode(@"//section[@class=""basic-info""]/div[@class=""active-dates""]/div"));
//genre
var genres = string.Empty;
var genreNodes = mainPage.DocumentNode.SelectNodes(@"//section[@class=""basic-info""]/div[@class=""genre""]/div/a");
if (genreNodes != null)
{
genres = genreNodes.Aggregate(genres, (current, genre) => current + (AllmusicSiteScraper.CleanInnerText(genre) + ", "));
genres = genres.TrimEnd(new[] { ',', ' ' }); // remove trailing ", "
}
// born / formed
var born = AllmusicSiteScraper.CleanInnerText(mainPage.DocumentNode.SelectSingleNode(@"//section[@class=""basic-info""]/div[@class=""birth""]/div"));
// styles
var styles = string.Empty;
var styleNodes = mainPage.DocumentNode.SelectNodes(@"//section[@class=""basic-info""]/div[@class=""styles""]/div/a");
if (styleNodes != null)
{
styles = styleNodes.Aggregate(styles, (current, style) => current + (AllmusicSiteScraper.CleanInnerText(style) + ", "));
styles = styles.TrimEnd(new[] { ',', ' ' }); // remove trailing ", "
}
// bio
var bio = string.Empty;
var bioURL = "http://www.allmusic.com/" + AllmusicSiteScraper.CleanAttribute(mainPage.DocumentNode.SelectSingleNode(@"//ul[@class=""tabs overview""]/li[@class=""tab biography""]/a"), "href");
if (!string.IsNullOrEmpty(bioURL))
{
var bioPage = new HtmlWeb().Load(bioURL);
bio = AllmusicSiteScraper.CleanInnerText(bioPage.DocumentNode.SelectSingleNode(@"//section[@class=""biography""]/div[@class=""text""]"));
}
// albums
var albumList = string.Empty;
var albumPageURL = "http://www.allmusic.com/" + AllmusicSiteScraper.CleanAttribute(mainPage.DocumentNode.SelectSingleNode(@"//ul[@class=""tabs overview""]/li[@class=""tab discography""]/a"), "href");
if (!string.IsNullOrEmpty(albumPageURL))
{
var albumPage = new HtmlWeb().Load(albumPageURL);
var albums = albumPage.DocumentNode.SelectNodes(@"//section[@class=""discography""]/table/tbody/tr");
if (albums != null)
{
foreach (var album in albums)
{
var year = AllmusicSiteScraper.CleanInnerText(album.SelectSingleNode(@"td[@class=""year""]"));
var title = AllmusicSiteScraper.CleanInnerText(album.SelectSingleNode(@"td[@class=""title""]/a"));
var label = AllmusicSiteScraper.CleanInnerText(album.SelectSingleNode(@"td[@class=""label""]"));
albumList += year + " - " + title + " (" + label + ")" + Environment.NewLine;
}
}
}
var artistInfo = new ArtistInfo
{
AMGBio = bio,
Albums = albumList,
Artist = artistName,
Born = born,
Compilations = string.Empty,
Genres = genres,
Image = artistImg,
Instruments = string.Empty,
Misc = string.Empty,
Singles = string.Empty,
Styles = styles,
Tones = moods,
YearsActive = yearsActive
};
Set(artistInfo);
_bLoaded = true;
return(_bLoaded);
}
/// <summary>
/// Take URL of an album details page and scrape details
/// </summary>
/// <param name="strUrl">URL of album details page</param>
/// <returns>True if scrape was successful</returns>
public bool Parse(string strUrl)
{
var albumPage = new HtmlWeb().Load(strUrl);
// artist
var strAlbumArtist = AllmusicSiteScraper.CleanInnerText(albumPage.DocumentNode.SelectSingleNode(@"//h3[@class=""album-artist""]/span/a"));
// album
var strAlbum = AllmusicSiteScraper.CleanInnerText(albumPage.DocumentNode.SelectSingleNode(@"//h2[@class=""album-title""]"));
// Image URL
var imgURL =
AllmusicSiteScraper.CleanAttribute(
albumPage.DocumentNode.SelectSingleNode(@"//div[@class=""album-cover""]/div[@class=""album-contain""]/img"),
"src");
// Rating
var iRating = 0;
var ratingMatch = AllmusicSiteScraper.CleanInnerText(albumPage.DocumentNode.SelectSingleNode(@"//div[starts-with(@class,""allmusic-rating rating-allmusic"")]"));
int.TryParse(ratingMatch, out iRating);
// year
var iYear = 0;
var yearMatch = AllmusicSiteScraper.CleanInnerText(albumPage.DocumentNode.SelectSingleNode(@"//div[@class=""release-date""]/span"));
yearMatch = Regex.Replace(yearMatch, @".*(\d{4})", @"$1");
int.TryParse(yearMatch, out iYear);
// review
var strReview = AllmusicSiteScraper.CleanInnerText(albumPage.DocumentNode.SelectSingleNode(@"//div[@itemprop=""reviewBody""]"));
// build up track listing into one string
var strTracks = string.Empty;
var trackNodes = albumPage.DocumentNode.SelectNodes(@"//tr[@itemprop=""track""]");
if (trackNodes != null)
{
foreach (var track in trackNodes)
{
var trackNo = AllmusicSiteScraper.CleanInnerText(track.SelectSingleNode(@"td[@class=""tracknum""]"));
var title =
AllmusicSiteScraper.CleanInnerText(
track.SelectSingleNode(@"td[@class=""title-composer""]/div[@class=""title""]/a"));
var strDuration = AllmusicSiteScraper.CleanInnerText(track.SelectSingleNode(@"td[@class=""time""]"));
var iDuration = 0;
var iPos = strDuration.IndexOf(":", StringComparison.Ordinal);
if (iPos >= 0)
{
var strMin = strDuration.Substring(0, iPos);
var strSec = strDuration.Substring(iPos + 1);
int iMin = 0, iSec = 0;
Int32.TryParse(strMin, out iMin);
Int32.TryParse(strSec, out iSec);
iDuration = (iMin * 60) + iSec;
}
strTracks += trackNo + "@" + title + "@" + iDuration.ToString(CultureInfo.InvariantCulture) + "|";
}
}
// genres
var strGenres = string.Empty;
var genreNodes = albumPage.DocumentNode.SelectNodes(@"//section[@class=""basic-info""]/div[@class=""genre""]/div/a");
if (genreNodes != null)
{
strGenres = genreNodes.Aggregate(strGenres, (current, genre) => current + (AllmusicSiteScraper.CleanInnerText(genre) + ", "));
strGenres = strGenres.TrimEnd(new[] { ',', ' ' }); // remove trailing ", "
}
// build up styles into one string
var strThemes = string.Empty;
var themeNodes = albumPage.DocumentNode.SelectNodes(@"//section[@class=""themes""]/div/span[@class=""theme""]/a");
if (themeNodes != null)
{
strThemes = themeNodes.Aggregate(strThemes, (current, theme) => current + (AllmusicSiteScraper.CleanInnerText(theme) + ", "));
strThemes = strThemes.TrimEnd(new[] { ',', ' ' }); // remove trailing ", "
}
// build up moods into one string
var strMoods = string.Empty;
var moodNodes = albumPage.DocumentNode.SelectNodes(@"//section[@class=""moods""]/div/span[@class=""mood""]/a");
if (moodNodes != null)
{
strMoods = moodNodes.Aggregate(strMoods, (current, mood) => current + (AllmusicSiteScraper.CleanInnerText(mood) + ", "));
strMoods = strMoods.TrimEnd(new[] { ',', ' ' }); // remove trailing ", "
}
var album = new AlbumInfo
{
Album = strAlbum,
Artist = strAlbumArtist,
Genre = strGenres,
Tones = strMoods,
Styles = strThemes,
Review = strReview,
Image = imgURL,
Rating = iRating,
Tracks = strTracks,
AlbumArtist = strAlbumArtist,
Year = iYear
};
Set(album);
Loaded = true;
return(true);
}
