public bool Parse(string html)
{
_songs.Clear();
HTMLUtil util = new HTMLUtil();
string strHtmlLow = html.ToLower();
int begIndex = 0;
int endIndex = 0;
// Extract Cover URL
string pattern = @"<!--Begin.*?Album.*?Photo-->\s*?.*?<img.*?src=\""(.*?)\""";
if (FindPattern(pattern, html))
{
_strImageURL = _match.Groups[1].Value;
}
// Extract Review
pattern = @"<td.*?class=""tab_off""><a.*?href=""(.*?)"">.*?Review.*?</a>";
if (FindPattern(pattern, html))
{
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(_match.Groups[1].Value);
pattern = @"<p.*?class=""author"">.*\s*?.*?<p.*?class=""text"">(.*?)</p>";
if (FindPattern(pattern, contentinfo))
{
string data = _match.Groups[1].Value;
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out data);
_strReview = data.Trim();
}
}
catch (Exception) {}
}
// Extract Artist
pattern = @"<h3.*?artist</h3>\s*?.*?<a.*"">(.*)</a>";
if (FindPattern(pattern, html))
{
_artist = _match.Groups[1].Value;
util.RemoveTags(ref _artist);
}
// Extract Album
pattern = @"<h3.*?album</h3>\s*?.*?<p>(.*)</P>";
if (FindPattern(pattern, html))
{
_strTitle = _match.Groups[1].Value;
util.RemoveTags(ref _strTitle);
}
// Extract Rating
pattern = @"<h3.*?rating</h3>\s*?.*?src=""(.*?)""";
if (FindPattern(pattern, html))
{
string strRating = _match.Groups[1].Value;
util.RemoveTags(ref strRating);
strRating = strRating.Substring(26, 1);
try
{
_iRating = Int32.Parse(strRating);
}
catch (Exception) {}
}
// Release Date
pattern = @"<h3.*?release.*?date</h3>\s*?.*?<p>(.*)</P>";
if (FindPattern(pattern, html))
{
_strDateOfRelease = _match.Groups[1].Value;
util.RemoveTags(ref _strDateOfRelease);
// extract the year out of something like "1998 (release)" or "12 feb 2003"
int nPos = _strDateOfRelease.IndexOf("19");
if (nPos > -1)
{
if ((int)_strDateOfRelease.Length >= nPos + 3 && Char.IsDigit(_strDateOfRelease[nPos + 2]) &&
Char.IsDigit(_strDateOfRelease[nPos + 3]))
{
string strYear = _strDateOfRelease.Substring(nPos, 4);
_strDateOfRelease = strYear;
}
else
{
nPos = _strDateOfRelease.IndexOf("19", nPos + 2);
if (nPos > -1)
{
if ((int)_strDateOfRelease.Length >= nPos + 3 && Char.IsDigit(_strDateOfRelease[nPos + 2]) &&
Char.IsDigit(_strDateOfRelease[nPos + 3]))
{
string strYear = _strDateOfRelease.Substring(nPos, 4);
_strDateOfRelease = strYear;
}
}
}
}
nPos = _strDateOfRelease.IndexOf("20");
if (nPos > -1)
{
if ((int)_strDateOfRelease.Length > nPos + 3 && Char.IsDigit(_strDateOfRelease[nPos + 2]) &&
Char.IsDigit(_strDateOfRelease[nPos + 3]))
{
string strYear = _strDateOfRelease.Substring(nPos, 4);
_strDateOfRelease = strYear;
}
else
{
nPos = _strDateOfRelease.IndexOf("20", nPos + 1);
if (nPos > -1)
{
if ((int)_strDateOfRelease.Length > nPos + 3 && Char.IsDigit(_strDateOfRelease[nPos + 2]) &&
Char.IsDigit(_strDateOfRelease[nPos + 3]))
{
string strYear = _strDateOfRelease.Substring(nPos, 4);
_strDateOfRelease = strYear;
}
}
}
}
}
// Extract Genre
begIndex = strHtmlLow.IndexOf("<h3>genre</h3>");
endIndex = strHtmlLow.IndexOf("</div>", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = html.Substring(begIndex, endIndex - begIndex);
pattern = @"(<li>(.*?)</li>)";
if (FindPattern(pattern, contentInfo))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strGenre);
_strGenre = _strGenre.Trim(new[] { ' ', ',' });
}
}
// Extract Styles
begIndex = strHtmlLow.IndexOf("<h3>style</h3>");
endIndex = strHtmlLow.IndexOf("</div>", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = html.Substring(begIndex, endIndex - begIndex);
pattern = @"(<li>(.*?)</li>)";
if (FindPattern(pattern, contentInfo))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strStyles);
_strStyles = _strStyles.Trim(new[] { ' ', ',' });
}
}
// Extract Moods
begIndex = strHtmlLow.IndexOf("<h3>moods</h3>");
endIndex = strHtmlLow.IndexOf("</div>", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = html.Substring(begIndex, endIndex - begIndex);
pattern = @"(<li>(.*?)</li>)";
if (FindPattern(pattern, contentInfo))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strTones);
_strTones = _strTones.Trim(new[] { ' ', ',' });
}
}
// Extract Songs
begIndex = strHtmlLow.IndexOf("<!-- tracks table -->");
endIndex = strHtmlLow.IndexOf("<!-- end tracks table -->", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = html.Substring(begIndex, endIndex - begIndex);
pattern = @"<tr.*class=""visible"".*?\s*?<td.*</td>\s*?.*<td.*</td>\s*?.*<td.*?>(?<track>.*)</td>" +
@"\s*?.*<td.*</td>\s*?.*<td.*?>(?<title>.*)</td>\s*?.*?<td.*?>\s*?.*</td>\s*?.*?<td.*?>(?<duration>.*)</td>";
if (FindPattern(pattern, contentInfo))
{
while (_match.Success)
{
// Tracknumber
int iTrack = 0;
try
{
iTrack = Int32.Parse(_match.Groups["track"].Value);
}
catch (Exception) {}
// Song Title
string strTitle = _match.Groups["title"].Value;
util.RemoveTags(ref strTitle);
util.ConvertHTMLToAnsi(strTitle, out strTitle);
// Duration
int iDuration = 0;
string strDuration = _match.Groups["duration"].Value;
int iPos = strDuration.IndexOf(":");
if (iPos >= 0)
{
string strMin, strSec;
strMin = strDuration.Substring(0, iPos);
iPos++;
strSec = strDuration.Substring(iPos);
int iMin = 0, iSec = 0;
try
{
iMin = Int32.Parse(strMin);
iSec = Int32.Parse(strSec);
}
catch (Exception) {}
iDuration = iMin * 60 + iSec;
}
// Create new song object
MusicSong newSong = new MusicSong();
newSong.Track = iTrack;
newSong.SongName = strTitle;
newSong.Duration = iDuration;
_songs.Add(newSong);
_match = _match.NextMatch();
}
}
}
// Set to "Not available" if no value from web
if (_artist.Length == 0)
{
_artist = GUILocalizeStrings.Get(416);
}
if (_strDateOfRelease.Length == 0)
{
_strDateOfRelease = GUILocalizeStrings.Get(416);
}
if (_strGenre.Length == 0)
{
_strGenre = GUILocalizeStrings.Get(416);
}
if (_strTones.Length == 0)
{
_strTones = GUILocalizeStrings.Get(416);
}
if (_strStyles.Length == 0)
{
_strStyles = GUILocalizeStrings.Get(416);
}
if (_strTitle.Length == 0)
{
_strTitle = GUILocalizeStrings.Get(416);
}
if (_strTitle2.Length == 0)
{
_strTitle2 = _strTitle;
}
Loaded = true;
return(true);
}
/// <summary>
/// Parse the Detail Page returned from the Allmusic Scraper
/// </summary>
/// <param name="strHTML"></param>
/// <returns></returns>
public bool Parse(string strHTML)
{
HTMLUtil util = new HTMLUtil();
int begIndex = 0;
int endIndex = 0;
string strHTMLLow = strHTML.ToLower();
// Get the Artist Name
string pattern = @"<h1.*class=""title"">(.*)</h1>";
if (!FindPattern(pattern, strHTML))
{
return(false);
}
_strArtistName = _match.Groups[1].Value;
// Born
pattern = @"<h3>.*Born.*</h3>\s*?<p>(.*)</p>";
if (FindPattern(pattern, strHTML))
{
string strValue = _match.Groups[1].Value;
util.RemoveTags(ref strValue);
util.ConvertHTMLToAnsi(strValue, out _strBorn);
_strBorn = _strBorn.Trim();
}
// Years Active
pattern = @"(<span.*?class=""active"">(.*?)</span>)";
if (FindPattern(pattern, strHTML))
{
while (_match.Success)
{
_strYearsActive += string.Format("{0}s, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
_strYearsActive = _strYearsActive.Trim(new[] { ' ', ',' });
}
// Genre
pattern = @"<div.*?id=""genre-style"">\s*?.*?\s*?<h3>.*?Genres.*?</h3>\s*?.*?(<p>(.*?)</p>)";
if (FindPattern(pattern, strHTML))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strGenres);
_strGenres = _strGenres.Trim(new[] { ' ', ',' });
}
// Style
begIndex = strHTMLLow.IndexOf("<h3>styles</h3>");
endIndex = strHTMLLow.IndexOf("<!--end genre/styles-->", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = strHTML.Substring(begIndex, endIndex - begIndex);
pattern = @"(<li>(.*?)</li>)";
if (FindPattern(pattern, contentInfo))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strStyles);
_strStyles = _strStyles.Trim(new[] { ' ', ',' });
}
}
// Mood
begIndex = strHTMLLow.IndexOf("<h3>moods</h3>");
endIndex = strHTMLLow.IndexOf("</div>", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = strHTML.Substring(begIndex, endIndex - begIndex);
pattern = @"(<li>(.*?)</li>)";
if (FindPattern(pattern, contentInfo))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strTones);
_strTones = _strTones.Trim(new[] { ' ', ',' });
}
}
// Instruments
begIndex = strHTMLLow.IndexOf("<h3>instruments</h3>");
endIndex = strHTMLLow.IndexOf("</div>", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
string contentInfo = strHTML.Substring(begIndex, endIndex - begIndex);
if (FindPattern(pattern, contentInfo))
{
string data = "";
while (_match.Success)
{
data += string.Format("{0}, ", _match.Groups[2].Value);
_match = _match.NextMatch();
}
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out _strInstruments);
_strInstruments = _strInstruments.Trim(new[] { ' ', ',' });
}
}
// picture URL
pattern = @"<div.*?class=""image"">\s*?.*<img.*id=""artist_image"".*?src=\""(.*?)\""";
if (FindPattern(pattern, strHTML))
{
_strArtistPictureURL = _match.Groups[1].Value;
}
// parse AMG BIOGRAPHY
pattern = @"<td.*?class=""tab_off""><a.*?href=""(.*?)"">.*?Biography.*?</a>";
if (FindPattern(pattern, strHTML))
{
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(_match.Groups[1].Value);
begIndex = contentinfo.IndexOf("<!--Begin Biography -->");
endIndex = contentinfo.IndexOf("</div>", begIndex + 2);
if (begIndex != -1 && endIndex != -1)
{
pattern = @"<p.*?class=""text"">(.*?)</p>";
if (FindPattern(pattern, contentinfo))
{
string data = _match.Groups[1].Value;
util.RemoveTags(ref data);
util.ConvertHTMLToAnsi(data, out data);
_strAMGBiography = data.Trim();
}
}
}
catch (Exception) {}
}
string compilationPage = "";
string singlesPage = "";
string dvdPage = "";
string miscPage = "";
// discography (albums)
pattern = @"<td.*class=""tab_off""><a.*?href=""(.*?)"">.*Discography.*</a>";
if (FindPattern(pattern, strHTML))
{
// Get Link to other sub pages
compilationPage = _match.Groups[1].Value + "/compilations";
singlesPage = _match.Groups[1].Value + "/singles-eps";
dvdPage = _match.Groups[1].Value + "/dvds-videos";
miscPage = _match.Groups[1].Value + "/other";
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(_match.Groups[1].Value);
pattern = @"sorted.*? cell"">(?<year>.*?)</td>\s*?.*?</td>\s*.*?<a.*?"">(?<album>.*?)" +
@"</a>.*?</td>\s*.*?</td>\s*.*?"">(?<label>.*?)</td>";
if (FindPattern(pattern, contentinfo))
{
while (_match.Success)
{
string year = _match.Groups["year"].Value;
string albumTitle = _match.Groups["album"].Value;
string label = _match.Groups["label"].Value;
util.RemoveTags(ref year);
util.ConvertHTMLToAnsi(year, out year);
util.RemoveTags(ref albumTitle);
util.ConvertHTMLToAnsi(albumTitle, out albumTitle);
util.RemoveTags(ref label);
util.ConvertHTMLToAnsi(label, out label);
try
{
string[] dAlbumInfo = { year.Trim(), albumTitle.Trim(), label.Trim() };
_discographyAlbum.Add(dAlbumInfo);
}
catch {}
_match = _match.NextMatch();
}
}
}
catch (Exception) {}
}
// Compilations
if (compilationPage != "")
{
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(compilationPage);
pattern = @"sorted.*? cell"">(?<year>.*?)</td>\s*?.*?</td>\s*.*?<a.*?"">(?<album>.*?)" +
@"</a>.*?</td>\s*.*?</td>\s*.*?"">(?<label>.*?)</td>";
if (FindPattern(pattern, contentinfo))
{
while (_match.Success)
{
string year = _match.Groups["year"].Value;
string albumTitle = _match.Groups["album"].Value;
string label = _match.Groups["label"].Value;
util.RemoveTags(ref year);
util.ConvertHTMLToAnsi(year, out year);
util.RemoveTags(ref albumTitle);
util.ConvertHTMLToAnsi(albumTitle, out albumTitle);
util.RemoveTags(ref label);
util.ConvertHTMLToAnsi(label, out label);
try
{
string[] dAlbumInfo = { year.Trim(), albumTitle.Trim(), label.Trim() };
_discographyCompilations.Add(dAlbumInfo);
}
catch {}
_match = _match.NextMatch();
}
}
}
catch (Exception) {}
}
// Singles
if (singlesPage != "")
{
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(singlesPage);
pattern = @"sorted.*? cell"">(?<year>.*?)</td>\s*?.*?</td>\s*.*?<a.*?"">(?<album>.*?)" +
@"</a>.*?</td>\s*.*?</td>\s*.*?"">(?<label>.*?)</td>";
if (FindPattern(pattern, contentinfo))
{
while (_match.Success)
{
string year = _match.Groups["year"].Value;
string albumTitle = _match.Groups["album"].Value;
string label = _match.Groups["label"].Value;
util.RemoveTags(ref year);
util.ConvertHTMLToAnsi(year, out year);
util.RemoveTags(ref albumTitle);
util.ConvertHTMLToAnsi(albumTitle, out albumTitle);
util.RemoveTags(ref label);
util.ConvertHTMLToAnsi(label, out label);
try
{
string[] dAlbumInfo = { year.Trim(), albumTitle.Trim(), label.Trim() };
_discographySingles.Add(dAlbumInfo);
}
catch {}
_match = _match.NextMatch();
}
}
}
catch (Exception) {}
}
// DVD Videos
if (dvdPage != "")
{
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(dvdPage);
pattern = @"sorted.*? cell"">(?<year>.*?)</td>\s*?.*?</td>\s*.*?<a.*?"">(?<album>.*?)" +
@"</a>.*?</td>\s*.*?</td>\s*.*?"">(?<label>.*?)</td>";
if (FindPattern(pattern, contentinfo))
{
while (_match.Success)
{
string year = _match.Groups["year"].Value;
string albumTitle = _match.Groups["album"].Value;
string label = _match.Groups["label"].Value;
util.RemoveTags(ref year);
util.ConvertHTMLToAnsi(year, out year);
util.RemoveTags(ref albumTitle);
util.ConvertHTMLToAnsi(albumTitle, out albumTitle);
util.RemoveTags(ref label);
util.ConvertHTMLToAnsi(label, out label);
try
{
string[] dAlbumInfo = { year.Trim(), albumTitle.Trim(), label.Trim() };
_discographyMisc.Add(dAlbumInfo);
}
catch {}
_match = _match.NextMatch();
}
}
}
catch (Exception) {}
}
// Other
if (miscPage != "")
{
try
{
string contentinfo = AllmusicSiteScraper.GetHTTP(miscPage);
pattern = @"sorted.*? cell"">(?<year>.*?)</td>\s*?.*?</td>\s*.*?<a.*?"">(?<album>.*?)" +
@"</a>.*?</td>\s*.*?</td>\s*.*?"">(?<label>.*?)</td>";
if (FindPattern(pattern, contentinfo))
{
while (_match.Success)
{
string year = _match.Groups["year"].Value;
string albumTitle = _match.Groups["album"].Value;
string label = _match.Groups["label"].Value;
util.RemoveTags(ref year);
util.ConvertHTMLToAnsi(year, out year);
util.RemoveTags(ref albumTitle);
util.ConvertHTMLToAnsi(albumTitle, out albumTitle);
util.RemoveTags(ref label);
util.ConvertHTMLToAnsi(label, out label);
try
{
string[] dAlbumInfo = { year.Trim(), albumTitle.Trim(), label.Trim() };
_discographyMisc.Add(dAlbumInfo);
}
catch {}
_match = _match.NextMatch();
}
}
}
catch (Exception) {}
}
_bLoaded = true;
return(_bLoaded);
}