/// <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;
}
public bool FindAlbuminfo(string strAlbum, string artistName, int releaseYear)
{
_albumList.Clear();
// strAlbum="1999";//escapolygy";
// make request
// type is
// http://www.allmusic.com/cg/amg.dll?P=amg&SQL=escapolygy&OPT1=2
HTMLUtil util = new HTMLUtil();
string postData = String.Format("P=amg&SQL={0}&OPT1=2", HttpUtility.UrlEncode(strAlbum));
string html = PostHTTP("http://www.allmusic.com/cg/amg.dll", postData);
if (html.Length == 0)
{
return false;
}
// check if this is an album
MusicAlbumInfo newAlbum = new MusicAlbumInfo();
newAlbum.AlbumURL = "http://www.allmusic.com/cg/amg.dll?" + postData;
if (newAlbum.Parse(html))
{
_albumList.Add(newAlbum);
return true;
}
string htmlLow = html;
htmlLow = htmlLow.ToLower();
int startOfTable = htmlLow.IndexOf("id=\"expansiontable1\"");
if (startOfTable < 0)
{
return false;
}
startOfTable = htmlLow.LastIndexOf("<table", startOfTable);
if (startOfTable < 0)
{
return false;
}
HTMLTable table = new HTMLTable();
string strTable = html.Substring(startOfTable);
table.Parse(strTable);
for (int i = 1; i < table.Rows; ++i)
{
HTMLTable.HTMLRow row = table.GetRow(i);
string albumName = "";
string albumUrl = "";
string nameOfAlbum = "";
string nameOfArtist = "";
for (int iCol = 0; iCol < row.Columns; ++iCol)
{
string column = row.GetColumValue(iCol);
if (iCol == 1 && (column.Length != 0))
{
albumName = "(" + column + ")";
}
if (iCol == 2)
{
nameOfArtist = column;
util.RemoveTags(ref nameOfArtist);
if (!column.Equals(" "))
{
albumName = String.Format("- {0} {1}", nameOfArtist, albumName);
}
}
if (iCol == 4)
{
string tempAlbum = column;
util.RemoveTags(ref tempAlbum);
albumName = String.Format("{0} {1}", tempAlbum, albumName);
nameOfAlbum = tempAlbum;
}
if (iCol == 4 && column.IndexOf("<a href=\"") >= 0)
{
int pos1 = column.IndexOf("<a href=\"");
pos1 += +"<a href=\"".Length;
int iPos2 = column.IndexOf("\">", pos1);
if (iPos2 >= 0)
{
if (nameOfAlbum.Length == 0)
{
nameOfAlbum = albumName;
}
// full album url:
// http://www.allmusic.com/cg/amg.dll?p=amg&token=&sql=10:66jieal64xs7
string url = column.Substring(pos1, iPos2 - pos1);
string albumNameStripped;
albumUrl = String.Format("http://www.allmusic.com{0}", url);
MusicAlbumInfo newAlbumInfo = new MusicAlbumInfo();
util.ConvertHTMLToAnsi(albumName, out albumNameStripped);
newAlbumInfo.Title2 = albumNameStripped;
newAlbumInfo.AlbumURL = util.ConvertHTMLToAnsi(albumUrl);
newAlbumInfo.Artist = util.ConvertHTMLToAnsi(nameOfArtist);
newAlbumInfo.Title = util.ConvertHTMLToAnsi(nameOfAlbum);
_albumList.Add(newAlbumInfo);
}
}
}
}
// now sort
_albumList.Sort(new AlbumSort(strAlbum, artistName, releaseYear));
return true;
}
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>
/// Search on Allmusic for the requested string
/// </summary>
/// <param name="searchBy"></param>
/// <param name="searchStr"></param>
/// <returns></returns>
public bool FindInfo(SearchBy searchBy, string searchStr)
{
_searchby = searchBy;
HTMLUtil util = new HTMLUtil();
string strPostData = "";
if (SearchBy.Albums == searchBy)
{
strPostData = string.Format(ALBUMSEARCH, HttpUtility.UrlEncode(searchStr));
}
else
{
searchStr = SwitchArtist(searchStr);
strPostData = string.Format(ARTISTSEARCH, HttpUtility.UrlEncode(searchStr));
}
string strHTML = PostHTTP(MAINURL + URLPROGRAM, strPostData);
if (strHTML.Length == 0)
{
return false;
}
_htmlCode = strHTML; // save the html content...
Regex multiples = new Regex(
@"\sSearch\sResults\sfor:",
RegexOptions.IgnoreCase
| RegexOptions.Multiline
| RegexOptions.IgnorePatternWhitespace
| RegexOptions.Compiled
);
if (multiples.IsMatch(strHTML))
{
string pattern = "";
if (searchBy.ToString().Equals("Artists"))
{
pattern = @"<tr.*?>\s*?.*?<td\s*?class=""relevance\stext-center"">\s*?.*\s*?.*</td>" +
@"\s*?.*<td.*\s*?.*</td>\s*?.*<td>.*<a.*href=""(?<code>.*?)"">(?<name>.*)</a>.*</td>" +
@"\s*?.*<td>(?<detail>.*)</td>\s*?.*<td>(?<detail2>.*)</td>";
}
else if (searchBy.ToString().Equals("Albums"))
{
pattern = @"<tr.*?>\s*?.*?<td\s*?class=""relevance\stext-center"">\s*?.*\s*?.*</td>" +
@"\s*?.*<td.*\s*?.*</td>\s*?.*<td>.*<a.*href=""(?<code>.*?)"">(?<name>.*)</a>.*</td>" +
@"\s*?.*<td>(?<detail>.*)</td>\s*?.*<td>.*</td>\s*?.*<td>(?<detail2>.*)</td>";
}
Match m;
Regex itemsFoundFromSite = new Regex(
pattern,
RegexOptions.IgnoreCase
| RegexOptions.Multiline
| RegexOptions.IgnorePatternWhitespace
| RegexOptions.Compiled
);
for (m = itemsFoundFromSite.Match(strHTML); m.Success; m = m.NextMatch())
{
string code = m.Groups["code"].ToString();
string name = m.Groups["name"].ToString();
string detail = m.Groups["detail"].ToString();
string detail2 = m.Groups["detail2"].ToString();
util.RemoveTags(ref name);
util.ConvertHTMLToAnsi(name, out name);
util.RemoveTags(ref detail);
util.ConvertHTMLToAnsi(detail, out detail);
util.RemoveTags(ref detail2);
util.ConvertHTMLToAnsi(detail2, out detail2);
if (SearchBy.Artists == searchBy)
{
detail += " - " + detail2;
if (detail.Length > 0)
{
_codes.Add(code);
_values.Add(name + " - " + detail);
}
else
{
_codes.Add(code);
_values.Add(name);
}
}
else
{
MusicAlbumInfo albumInfo = new MusicAlbumInfo();
albumInfo.AlbumURL = code;
albumInfo.Artist = detail;
albumInfo.Title = name;
albumInfo.DateOfRelease = detail2;
_albumList.Add(albumInfo);
}
}
_multiple = true;
}
else // found the right one
{}
return true;
}
public static string FindWithAction(string body, string keyStart, string keyEnd, string param1, string param2, string param3, string maxItems, string languages, out string allNames, out string allRoles, bool grabActorRoles)
{
allNames = string.Empty;
allRoles = string.Empty;
int iStart = 0;
int iEnd = 0;
int iLength = 0;
int maxItemsToAdd = 999; // Max number of items to add to matchgroup
if (!string.IsNullOrEmpty(maxItems)) maxItemsToAdd = Convert.ToInt32(maxItems);
string strTemp = String.Empty;
var htmlUtil = new HTMLUtil();
bool bregexs = false;
bool bregexe = false;
if (keyStart.StartsWith("#REGEX#"))
bregexs = true;
if (keyEnd.StartsWith("#REGEX#"))
bregexe = true;
if (keyStart != "" && keyEnd != "")
{
iLength = keyStart.Length;
if (param1.StartsWith("#REVERSE#"))
{
iStart = bregexs ? FindRegEx(body, keyStart, iStart, ref iLength, false) : body.LastIndexOf(keyStart);
}
else
if (bregexs)
iStart = FindRegEx(body, keyStart, iStart, ref iLength, true);
else
iStart = body.IndexOf(keyStart);
if (iStart > 0)
{
if (param1.StartsWith("#REVERSE#"))
{
iLength = keyEnd.Length;
if (bregexe)
iEnd = FindRegEx(body, keyEnd, iStart, ref iLength, false) + iStart;
else
iEnd = body.LastIndexOf(keyEnd, iStart);
}
else
{
iStart += iLength;
if (bregexe)
iEnd = FindRegEx(body, keyEnd, iStart, ref iLength, true) + iStart;
else
iEnd = body.IndexOf(keyEnd, iStart);
}
if (iEnd > 0)
{
if (param1.StartsWith("#REVERSE#"))
{
param1 = param1.Substring(9);
iEnd += iLength;
strTemp = body.Substring(iEnd, iStart - iEnd);
}
else
strTemp = body.Substring(iStart, iEnd - iStart);
if (strTemp != "")
{
//if (param3.Length > 0)
//{
// Regex oRegex = new Regex(param3);
// Regex oRegexReplace = new Regex(string.Empty);
// System.Text.RegularExpressions.MatchCollection oMatches = oRegex.Matches(strTemp);
// foreach (System.Text.RegularExpressions.Match oMatch in oMatches)
// {
// if (param1.StartsWith("#REGEX#"))
// {
// oRegexReplace = new Regex(param1.Substring(7));
// strTemp = strTemp.Replace(oMatch.Value, oRegexReplace.Replace(oMatch.Value, param2));
// }
// else
// strTemp = strTemp.Replace(param1, param2);
// }
//}
//else
//{
// if (param1.StartsWith("#REGEX#"))
// strTemp = Regex.Replace(strTemp, param1.Substring(7), param2);
// else
// if (param1.Length > 0)
// strTemp = strTemp.Replace(param1, param2);
//}
if (param3.Length > 0)
{
RegexOptions regexoption = new RegexOptions();
regexoption = RegexOptions.Singleline;
if (param3.StartsWith("#MULTILINE#"))
{
regexoption = RegexOptions.Multiline;
param3 = param3.Substring(11);
}
//Regex oRegex = new Regex(param3, RegexOptions.Singleline);
Regex oRegex = new Regex(param3, regexoption);
Regex oRegexReplace = new Regex(string.Empty);
strTemp = HttpUtility.HtmlDecode(strTemp);
if (regexoption != RegexOptions.Multiline)
strTemp = HttpUtility.HtmlDecode(strTemp).Replace("\n", "");
// System.Windows.Forms.Clipboard.SetDataObject(strTemp, false); // Must not be set when called by AMCupdater -> STAThread exception !
MatchCollection oMatches = oRegex.Matches(strTemp);
string strPerson = string.Empty;
string strRole = string.Empty;
if (oMatches.Count > 0)
{
string strCastDetails = "";
int i = 0;
foreach (System.Text.RegularExpressions.Match oMatch in oMatches)
{
strPerson = oMatch.Groups["person"].Value;
strPerson = Utils.stripHTMLtags(strPerson).Trim().Replace("\n", "");
//strPerson = HttpUtility.HtmlDecode(strPerson).Replace(",", ";");
strRole = oMatch.Groups["role"].Value;
strRole = Utils.stripHTMLtags(strRole).Trim().Replace("\n", "");
//strRole = HttpUtility.HtmlDecode(strRole).Replace(",", ";");
if (param1.Length > 0)
{
if (!string.IsNullOrEmpty(strPerson))
strPerson = ReplaceNormalOrRegex(strPerson, param1, "");
if (!string.IsNullOrEmpty(strRole))
strRole = ReplaceNormalOrRegex(strRole, param1, "");
//if (param1.StartsWith("#REGEX#"))
//{
// oRegexReplace = new Regex(param1.Substring(7));
// if (!string.IsNullOrEmpty(strPerson))
// strPerson = strPerson.Replace(strPerson, oRegexReplace.Replace(strPerson, param2)).Trim();
// if (!string.IsNullOrEmpty(strRole))
// strRole = strRole.Replace(strRole, oRegexReplace.Replace(strRole, param2)).Trim();
//}
//else
//{
// if (!string.IsNullOrEmpty(strActor))
// strActor = strActor.Replace(param1, param2).Trim();
// if (!string.IsNullOrEmpty(strRole))
// strRole = strRole.Replace(param1, param2).Trim();
//}
}
// build allNames & allRoles strings for dropdowns
if (!string.IsNullOrEmpty(allNames)) allNames += ", ";
allNames += strPerson;
if (!string.IsNullOrEmpty(allRoles)) allRoles += ", ";
allRoles += strRole;
if (i < maxItemsToAdd) // Limit number of items to add
{
string[] langSplit = languages.Split(new Char[] { ',', ';' }, StringSplitOptions.RemoveEmptyEntries); // language filter (grabber script or override)
if (string.IsNullOrEmpty(languages))
{
if (!string.IsNullOrEmpty(strCastDetails)) strCastDetails += ", ";
strCastDetails += strPerson;
if (strRole != string.Empty && grabActorRoles)
strCastDetails += " (" + strRole + ")";
//strCastDetails += "\n";
i = i + 1;
}
else
{
foreach (var s in langSplit)
{
if (strPerson.ToLower().Contains(s.Trim().ToLower()) || strRole.ToLower().Contains(s.Trim().ToLower()))
{
if (!string.IsNullOrEmpty(strCastDetails)) strCastDetails += ", ";
strCastDetails += strPerson;
if (strRole != string.Empty && grabActorRoles)
strCastDetails += " (" + strRole + ")";
// Don't add groupnames when adding TTitles
//strCastDetails += "\n";
i = i + 1;
}
}
strCastDetails = ReplaceNormalOrRegex(strCastDetails, param1, "");
//if (param1.StartsWith("#REGEX#"))
// strCastDetails = Regex.Replace(strCastDetails, param1.Substring(7), param2);
//else if (param1.Length > 0)
// strCastDetails = strCastDetails.Replace(param1, param2);
}
}
}
strTemp = strCastDetails;
if (param2.StartsWith("#")) strTemp = ReplaceNormalOrRegex(strTemp, param2, ""); // Cleanup only - no replacement of inner regex values
}
else // no matchcollection found
{
// strTemp = param2.StartsWith("#") ? ReplaceNormalOrRegex(strTemp, param2, "") : ReplaceNormalOrRegex(strTemp, param1, param2);
strTemp = ""; // set output to empty string, as no matches were found
}
string[] split = allNames.Split(new Char[] { ',', ';', '/' }, StringSplitOptions.RemoveEmptyEntries);
string strT = string.Empty;
string strTname = string.Empty;
string strTrole = string.Empty;
foreach (var str in split)
{
strT = str.Trim();
if (!string.IsNullOrEmpty(strTname)) strTname += ", ";
strTname += strT;
}
allNames = strTname;
split = allRoles.Split(new Char[] { ',', ';', '/' }, StringSplitOptions.RemoveEmptyEntries);
strT = string.Empty;
strTname = string.Empty;
strTrole = string.Empty;
foreach (var str in split)
{
strT = str.Trim();
if (!string.IsNullOrEmpty(strTrole)) strTrole += ", ";
strTrole += strT;
}
allRoles = strTrole;
}
else
{
strTemp = ReplaceNormalOrRegex(strTemp, param1, param2);
//System.Windows.Forms.Clipboard.SetDataObject(strTemp, false); // Must NOT be called, when using from AMCupdater, cause it's giving STAThread error !!!
}
// Added to enable the addition of surrounding chars for results...
if (param1.StartsWith("#ADD#") && param2.Length > 0)
{
string strExtend = param2;
string[] strExtendSplit = strExtend.Split(new Char[] { '|' }, StringSplitOptions.None); // { ',', ';', '|' }
if (strExtendSplit.Length == 2)
{
strTemp = strExtendSplit[0] + strTemp.Trim() + strExtendSplit[1];
}
else
{
strTemp = strExtend + strTemp.Trim() + strExtend;
}
}
strTemp = strTemp.Replace("„", "'").Replace("“", "'");
htmlUtil.RemoveTags(ref strTemp);
htmlUtil.ConvertHTMLToAnsi(strTemp, out strTemp);
// strTemp = System.Security.SecurityElement.Escape(strTemp); // escape invalid chars for valid XML value generation
// strTemp = strTemp.Replace("&", "&").Replace("<", "<").Replace(">", ">").Replace("\"", """).Replace("'", "'");
// strTemp = strTemp.Replace("<", "<").Replace(">", ">").Replace("\"", """).Replace("'", "'");
}
}
}
}
return strTemp.Trim();
}
public static string Find(string body, string keyStart, ref int iStart, string keyEnd)
{
int iEnd = 0;
int iLength = 0;
string strTemp = String.Empty;
var htmlUtil = new HTMLUtil();
bool bregexs = false;
bool bregexe = false;
if (keyStart.StartsWith("#REGEX#"))
bregexs = true;
if (keyEnd.StartsWith("#REGEX#"))
bregexe = true;
if (keyStart != "" && keyEnd != "")
{
iLength = keyStart.Length;
if (bregexs)
iStart = FindRegEx(body, keyStart, iStart, ref iLength, true) + iStart;
else
iStart = body.IndexOf(keyStart, iStart);
if (iStart >= 0)
{
iStart += iLength;
iLength = keyEnd.Length;
if (bregexe)
iEnd = FindRegEx(body, keyEnd, iStart, ref iLength, true) + iStart;
else
iEnd = body.IndexOf(keyEnd, iStart);
if (iEnd > 0)
{
strTemp = body.Substring(iStart, iEnd - iStart);
if (strTemp != "")
{
htmlUtil.RemoveTags(ref strTemp);
htmlUtil.ConvertHTMLToAnsi(strTemp, out strTemp);
}
}
}
}
return strTemp.Trim();
}
public static string FindWithAction(string body, string keyStart, ref int iStart, string keyEnd, string param1, string param2, string param3)
{
int iEnd = 0;
int iLength = 0;
var strTemp = String.Empty;
var htmlUtil = new HTMLUtil();
bool bregexs = false;
bool bregexe = false;
if (keyStart.StartsWith("#REGEX#"))
bregexs = true;
if (keyEnd.StartsWith("#REGEX#"))
bregexe = true;
if (keyStart != "" && keyEnd != "")
{
iLength = keyStart.Length;
if (bregexs)
iStart = FindRegEx(body, keyStart, iStart, ref iLength, true) + iStart;
else
iStart = body.IndexOf(keyStart, iStart);
if (iStart > 0)
{
iStart += iLength;
if (bregexe)
iEnd = FindRegEx(body, keyEnd, iStart, ref iLength, true) + iStart;
else
iEnd = body.IndexOf(keyEnd, iStart);
if (iEnd > 0)
{
strTemp = body.Substring(iStart, iEnd - iStart);
if (strTemp != "")
{
if (param3.Length > 0)
{
var oRegex = new Regex(param3);
var oRegexReplace = new Regex(string.Empty);
MatchCollection oMatches = oRegex.Matches(strTemp);
foreach (Match oMatch in oMatches)
{
if (param1.StartsWith("#REGEX#"))
{
oRegexReplace = new Regex(param1.Substring(7));
strTemp = strTemp.Replace(oMatch.Value, oRegexReplace.Replace(oMatch.Value, param2));
}
else
strTemp = strTemp.Replace(param1, param2);
}
}
else
{
strTemp = ReplaceNormalOrRegex(strTemp, param1, param2);
}
// Added to enable the addition of surrounding chars for results...
if (param1.StartsWith("#ADD#") && param2.Length > 0)
{
string strExtend = param2;
string[] strExtendSplit = strExtend.Split(new Char[] { '|' }, StringSplitOptions.None);
if (strExtendSplit.Length == 2)
{
strTemp = strExtendSplit[0] + strTemp.Trim() + strExtendSplit[1];
}
else
{
strTemp = strExtend + strTemp.Trim() + strExtend;
}
}
strTemp = strTemp.Replace("„", "'").Replace("“", "'");
htmlUtil.RemoveTags(ref strTemp);
htmlUtil.ConvertHTMLToAnsi(strTemp, out strTemp);
// strTemp = strTemp.Replace("<", "<").Replace(">", ">").Replace("\"", """).Replace("'", "'");
}
}
}
}
return strTemp.Trim();
}
