public static Song FetchSong(ScrapingBrowser Browser, songlist albumData, Guid songGuid, CrawlDatabaseEntities context)
{
Song song = new Song();
var title = albumData.songs[songGuid].song;
var slug = albumData.songs[songGuid].slug;
var artistGuid = albumData.songs[songGuid].artistId;
var endPathSongUrl = albumData.songs[songGuid].url;
var songSId = albumData.sid[songGuid];
var subDomain = (Convert.ToInt32(songSId, 16) - 100) / 7;
var songUrl = "http://mus" + subDomain + "." + TOPDOMAIN + endPathSongUrl;
var artistRouteName = albumData.artists[artistGuid].slug;
var albumRouteName = albumData.songs[songGuid].albumSlug;
var displaySongImage = "";
if (albumData.songs[songGuid].haveAlbumImage == "True")
{
displaySongImage = GetAlbumImage(artistRouteName, albumRouteName, 2);
}
else
{
displaySongImage = GetArtistImage(artistRouteName, 2);
}
// Get lyric
var lyricUrl = DOMAIN + String.Format("ajax/song?t=1&songId={0}", songGuid);
WebPage PageResult = Browser.NavigateToPage(new Uri(lyricUrl));
string lyrics = "";
ajax_response response = JsonConvert.DeserializeObject <ajax_response>(PageResult.ToString());
if (response.code == 0 && response.data != null && response.data.Count > 0)
{
lyrics = response.data[0].lyrics;
}
song.Guid = songGuid.ToString();
song.Title = title;
song.ArtistGuid = artistGuid.ToString();
song.MediaUrl = songUrl;
song.Thumbnail = displaySongImage;
song.Url = slug;
song.Lyrics = lyrics;
song.AlbumSlug = albumRouteName;
song.ArtistSlug = artistRouteName;
//bool hasSong = context.Songs.Where(s => s.Guid == song.Guid).Any();
//if (!hasSong)
//{
context.Songs.Add(song);
//}
// Save resources
//SaveImage(folderImagePath, songUrl);
//SaveImage(folderSongPath, songUrl);
return(song);
}
public static Album ScrapingAlbum(string link)
{
ScrapingBrowser Browser = new ScrapingBrowser();
Browser.AllowAutoRedirect = true; // Browser has settings you can access in setup
Browser.AllowMetaRedirect = true;
Browser.Encoding = Encoding.UTF8;
WebPage PageResult = Browser.NavigateToPage(new Uri(link));
var scriptString = PageResult.Html.CssSelect("#music>script").FirstOrDefault();
var engine = new Jurassic.ScriptEngine();
var result = engine.Evaluate("(function() { var MN = {};MN.m_page= {};MN.m_page.songlist = {};MN.m_page.songlist.artists = {};MN.m_page.songlist.songs = {};MN.m_page.songlist.sid = {};" + scriptString.InnerHtml + " return MN.m_page.songlist; })()");
var json = JSONObject.Stringify(engine, result);
songlist data = JsonConvert.DeserializeObject <songlist>(json);
// Get list song data of album
List <Guid> listSongGuid = new List <Guid>(data.songs.Keys);
List <Song> listSongs = new List <Song>();
for (int i = 0; i < listSongGuid.Count; i++)
{
listSongs.Add(FetchSong(Browser, data, listSongGuid[i]));
}
// Crawl album
Album album = new Album();
album.Title = PageResult.Html.CssSelect("#artist-info>article>header>h1").FirstOrDefault().InnerText.Trim();
album.ReleaseDate = PageResult.Html.CssSelect("#artist-info>article>header>h1>time").FirstOrDefault().InnerText.TrimStart('-').Trim();
album.ArtistName = PageResult.Html.CssSelect("#artist-info>article>header>h2>a").FirstOrDefault().InnerText.Trim();
album.Thumbnail = PageResult.Html.CssSelect("#artist-info>article>figure>img").FirstOrDefault().GetAttributeValue("src");
album.Slug = listSongs.Count > 0 ? listSongs[0].AlbumSlug: "";
SaveImage(folderAlbumImagePath, album.Thumbnail);
// Crawl artist
List <Artist> artist = new List <Artist>();
List <Guid> listArtistIds = new List <Guid>(data.artists.Keys);
for (int i = 0; i < listArtistIds.Count; i++)
{
var newArtist = new Artist()
{
Guid = listArtistIds[i],
Name = data.artists[listArtistIds[i]].artist,
Slug = data.artists[listArtistIds[i]].slug
};
artist.Add(newArtist);
var artistImageUrl = GetArtistImage(newArtist.Slug, 2);
SaveImage(folderArtistImagePath, artistImageUrl);
}
// Save mp3 files
for (int i = 0; i < listSongs.Count; i++)
{
var mp3FullPath = Path.Combine(folderSongPath, Path.GetFileName(listSongs[i].MediaUrl));
var success = FileDownloader.DownloadFile(listSongs[i].MediaUrl, mp3FullPath, 120000);
Console.WriteLine("Done - success: " + success);
}
return(null);
}
public static Album ScrapingAlbum(ScrapingBrowser Browser, string link)
{
CrawlDatabaseEntities context = new CrawlDatabaseEntities();
WebPage PageResult = Browser.NavigateToPage(new Uri(link));
var albumScript = PageResult.Html.CssSelect("head>script");
string scripts = "";
foreach (var item in albumScript)
{
if (!String.IsNullOrEmpty(item.GetAttributeValue("src")))
{
scripts += item.InnerHtml;
}
}
var scriptString = PageResult.Html.CssSelect("#music>script").FirstOrDefault();
var engine = new Jurassic.ScriptEngine();
var result = engine.Evaluate("(function() { var MN = {};MN.m_page= {};MN.m_page.songlist = {};MN.m_page.songlist.artists = {};MN.m_page.songlist.songs = {};MN.m_page.songlist.sid = {};" + scriptString.InnerHtml + " return MN.m_page.songlist; })()");
var json = JSONObject.Stringify(engine, result);
songlist data = JsonConvert.DeserializeObject <songlist>(json);
// Get list song data of album
List <Guid> listSongGuid = new List <Guid>(data.songs.Keys);
List <Song> listSongs = new List <Song>();
for (int i = 0; i < listSongGuid.Count; i++)
{
Console.WriteLine("------------- Begin song " + i);
Console.WriteLine("------------- Song link " + listSongGuid[i]);
listSongs.Add(FetchSong(Browser, data, listSongGuid[i], context));
Console.WriteLine("------------- End song " + i);
}
// Crawl album
Album album = new Album();
album.Title = PageResult.Html.CssSelect("#artist-info>article>header>h1").FirstOrDefault().InnerText.Trim();
album.ReleaseDate = PageResult.Html.CssSelect("#artist-info>article>header>h1>time").FirstOrDefault().InnerText.TrimStart('-').Trim();
album.ArtistName = PageResult.Html.CssSelect("#artist-info>article>header>h2>a").FirstOrDefault().InnerText.Trim();
album.Thumbnail = PageResult.Html.CssSelect("#artist-info>article>figure>img").FirstOrDefault().GetAttributeValue("src");
album.Slug = listSongs.Count > 0 ? listSongs[0].AlbumSlug: "";
album.Songs = listSongs;
SaveImage(folderAlbumImagePath, album.Thumbnail);
// Crawl artist
List <Artist> artist = new List <Artist>();
List <Guid> listArtistIds = new List <Guid>(data.artists.Keys);
album.ArtistGuid = listArtistIds.FirstOrDefault().ToString();
for (int i = 0; i < listArtistIds.Count; i++)
{
var artistImageUrl = GetArtistImage(data.artists[listArtistIds[i]].slug, 2);
var newArtist = new Artist()
{
Guid = listArtistIds[i].ToString(),
Name = data.artists[listArtistIds[i]].artist,
Slug = data.artists[listArtistIds[i]].slug,
Thumbnail = artistImageUrl
};
artist.Add(newArtist);
bool isArtistExisted = context.Artists.Where(a => a.Guid == newArtist.Guid).Any();
if (!isArtistExisted)
{
context.Artists.Add(newArtist);
SaveImage(folderArtistImagePath, artistImageUrl);
}
}
bool isAlbumExisted = context.Albums.Where(a => a.Title == album.Title && a.ArtistName == album.ArtistName).Any();
if (!isAlbumExisted)
{
context.Albums.Add(album);
}
context.SaveChanges();
// Save mp3 files
//for (int i = 0; i < listSongs.Count; i++)
//{
// var mp3FullPath = Path.Combine(folderSongPath, Path.GetFileName(listSongs[i].MediaUrl));
// var success = FileDownloader.DownloadFile(listSongs[i].MediaUrl, mp3FullPath, 120000);
// Console.WriteLine("Done - success: " + success);
//}
return(null);
}