private void ParseInfoBox(ImportedAlbumDataContract data, HtmlNode infoBox)
{
var statusRows = infoBox.SelectNodes("//p[@class='albumstatus']");
var artistRow = GetInfoElem(statusRows, "Artist");
if (artistRow != null)
{
var links = artistRow.SelectNodes("a");
data.ArtistNames = links.Select(l => l.InnerText).ToArray();
}
var releaseDateRow = GetInfoElem(statusRows, "Release");
if (releaseDateRow != null)
{
DateTime releaseDate;
if (DateTime.TryParseExact(releaseDateRow.Element("#text").InnerText, "yyyy.MM.dd", null, DateTimeStyles.None, out releaseDate))
{
data.ReleaseYear = releaseDate.Year;
}
}
var charaRow = GetInfoElem(statusRows, "Characters");
if (charaRow != null)
{
var charaImgs = charaRow.SelectNodes("a/img");
data.VocalistNames = charaImgs.Select(l => GetVocalistName(l.Attributes["src"].Value)).Where(l => l != null).ToArray();
}
}
public void SetUp()
{
_karenTDoc = ResourceHelper.ReadHtmlDocument("KarenT_SystemindParadox.htm");
_importer = new KarenTAlbumImporter(new PictureDownloaderStub());
_importedAlbum = _importer.GetAlbumData(_karenTDoc, "http://");
_importedData = _importedAlbum.Data;
}
public MikuDbAlbumContract GetAlbumData(HtmlDocument doc, string url)
{
var data = new ImportedAlbumDataContract();
var titleElem = doc.DocumentNode.SelectSingleNode("//div[@class = 'pgtitle_in']/h1/span");
if (titleElem != null)
{
data.Title = HtmlEntity.DeEntitize(titleElem.InnerText);
}
var mainPanel = doc.GetElementbyId("main_ref");
if (mainPanel != null)
{
var descBox = mainPanel.SelectSingleNode("p[@class = 'overview']");
if (descBox != null)
{
data.Description = descBox.InnerText;
}
var infoBox = mainPanel.SelectSingleNode("div[1]");
if (infoBox != null)
{
ParseInfoBox(data, infoBox);
}
var tracklistElem = mainPanel.SelectSingleNode("div[@class = 'songlistbox']");
if (tracklistElem != null)
{
ParseTracklist(data, tracklistElem);
}
}
var coverElem = doc.DocumentNode.SelectSingleNode("//div[@id = 'sub_ref']/div[@class = 'artwork']/div/a/img");
PictureDataContract coverPic = null;
if (coverElem != null)
{
coverPic = DownloadCoverPicture("https://karent.jp" + coverElem.Attributes["src"].Value);
}
return(new MikuDbAlbumContract(data)
{
CoverPicture = coverPic, SourceUrl = url
});
}
public MikuDbAlbumContract Parse(Stream input)
{
var tracks = new List <ImportedAlbumTrack>();
var parser = new DataRowParser();
var data = new ImportedAlbumDataContract();
data.Title = "Unknown";
using (var reader = new StreamReader(input)) {
string row;
while ((row = reader.ReadLine()) != null)
{
if (!parser.IsConfigured)
{
parser.Configure(row);
}
else
{
var dataRow = new DataRow(parser, row);
var albumName = dataRow.GetString(AlbumFileField.Album, string.Empty);
if (albumName != string.Empty)
{
data.Title = albumName;
}
var year = dataRow.GetIntOrDefault(AlbumFileField.Year, 0);
if (year != 0)
{
data.ReleaseYear = year;
}
var track = ParseTrack(dataRow, tracks.Count + 1);
tracks.Add(track);
}
}
}
data.ArtistNames = tracks.SelectMany(t => t.ArtistNames).Distinct().ToArray();
data.VocalistNames = tracks.SelectMany(t => t.VocalistNames).Distinct().ToArray();
data.Tracks = tracks.OrderBy(t => t.TrackNum).ToArray();
return(new MikuDbAlbumContract(data));
}
private void ParseTracklist(ImportedAlbumDataContract data, HtmlNode tracklistElem)
{
var songElems = tracklistElem.SelectNodes("//div[@class = 'song']");
var tracks = new List <ImportedAlbumTrack>();
for (int i = 1; i <= songElems.Count; ++i)
{
var songLink = songElems[i - 1].Element("a");
var track = ParseTrackRow(i, songLink.InnerText);
if (track != null)
{
tracks.Add(track);
}
}
data.Tracks = tracks.ToArray();
}
private MikuDbAlbumContract GetAlbumData(HtmlDocument doc, string url)
{
var data = new ImportedAlbumDataContract();
string title = string.Empty;
var titleElem = doc.DocumentNode.SelectSingleNode(".//h2[@class='posttitle']/a");
if (titleElem != null)
{
title = HtmlEntity.DeEntitize(titleElem.InnerText);
}
var coverPicLink = doc.DocumentNode.SelectSingleNode(".//div[@class='postcontent']/table/tr[1]/td[1]/a/img");
PictureDataContract coverPicture = null;
if (coverPicLink != null)
{
var address = coverPicLink.Attributes["src"].Value;
coverPicture = DownloadCoverPicture(address);
}
var infoBox = doc.DocumentNode.SelectSingleNode(".//div[@class='postcontent']/table/tr[1]/td[2]");
if (infoBox != null)
{
ParseInfoBox(data, infoBox);
}
var trackListRow = FindTracklistRow(doc, (infoBox != null ? infoBox.ParentNode.NextSibling : null));
if (trackListRow != null)
{
ParseTrackList(data, trackListRow);
}
return(new MikuDbAlbumContract {
Title = title, Data = data, CoverPicture = coverPicture, SourceUrl = url
});
}
private void ParseTrackList(ImportedAlbumDataContract data, HtmlNode cell)
{
var lines = cell.InnerText.Split(new[] { "<br>", "<br />", "\n" }, StringSplitOptions.RemoveEmptyEntries);
int discNum = 1;
var tracks = new List <ImportedAlbumTrack>();
foreach (var line in lines)
{
var dotPos = line.IndexOf('.');
if (dotPos <= 0)
{
continue;
}
var trackText = line.Substring(0, dotPos);
if (int.TryParse(trackText, out int trackNum))
{
if (trackNum == 1 && tracks.Any())
{
discNum++;
}
var trackTitle = line.Substring(dotPos + 1, line.Length - dotPos - 1).Trim();
trackTitle = trackTitle.Replace("(lyrics)", string.Empty);
tracks.Add(new ImportedAlbumTrack
{
DiscNum = discNum,
Title = HtmlEntity.DeEntitize(trackTitle),
TrackNum = trackNum
});
}
}
data.Tracks = tracks.ToArray();
}
private void ParseInfoBox(ImportedAlbumDataContract data, HtmlNode infoBox)
{
var text = infoBox.InnerHtml;
var rows = text.Split(new[] { "<br>", "<br />", "\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (var row in rows)
{
var stripped = HtmlEntity.DeEntitize(StripHtml(row));
if (stripped.StartsWith("\"") && stripped.EndsWith("\""))
{
stripped = stripped.Substring(1, stripped.Length - 2).Trim();
}
if (LineMatch(stripped, "Artist") || LineMatch(stripped, "Artists"))
{
var artists = ParseArtists(stripped.Substring(8));
data.ArtistNames = artists;
}
else if (LineMatch(stripped, "Vocals"))
{
var vocals = ParseArtists(stripped.Substring(8));
data.VocalistNames = vocals;
}
else if (LineMatch(stripped, "Circle"))
{
var artists = ParseArtist(stripped.Substring(8));
data.CircleName = artists;
}
else if (LineMatch(stripped, "Year"))
{
int year;
if (int.TryParse(stripped.Substring(6), out year))
{
data.ReleaseYear = year;
}
}
}
}
