{http://www.bootlegzone.com}

// ===================================================================
// ======== Godfather/bootlegzone.com auto-ID3 tagging script ========
// ===================================================================
// ========     Script I made to grab disc information        ========
// ========     and import ID3 tag info from bootlegzone      ========
// ========     disc pages and auto-import to MP3 files.      ========
// ========     Free to use/modify/distribute.                ========
// ========     Enjoy!                                        ========
// ========                                                   ========
// ========     Latest version can always be found here:      ========
// ========     http://www.pharaohweb.com/blog/?p=777         ========
// ========                                                   ========
// ========     dave                                          ========
// ===================================================================

Program bootlegzone;

var
  slMain: TStringList;
  iRow: integer;

  //---
  procedure GetArtistAlbum( sLine: string );
  var
    iPos: integer;
  begin
    sLine := on_cleanHTMLLine( sLine );
    iPos := rPos( ' -- ', sLine );
    if iPos > 0 then begin
       on_setArtist( Trim( Copy( sLine, 14, iPos - 14 ) ) );
       on_setAlbum( Trim( Copy( sLine, iPos + 3, 9999 ) ) );
    end;
  end;


  //---
  procedure GetImage( sLine: string );
  var
    iPos: integer;
    sTmp: string;
  begin
    iPos := Pos( '/scans/', sLine );
    if iPos > 0 then begin
       sTmp := Copy( sLine, iPos + 7, 99999 );
       iPos := Pos( '.jpg"', sTmp );
       if iPos = 0 then iPos := Pos( '.gif"', sTmp ); // maybe gif file
       if iPos > 0 then begin
          sTmp := Copy( sTmp, 1, iPos + 3 );
          sTmp := 'http://www.bootlegzone.com/beatleg/discs/scans/' + sTmp;
          on_setPicture( sTmp );
       end;
    end;
  end;

//=============================================================================================
//=======================       BEGIN track title super-loop BEGIN      =======================
//=============================================================================================

  procedure GetTracks( sLine: string );

  var
    iUpToRow, iPos, iPos2, nxtTrackLn, nxtArtistLn, medleyLoopRow, sameCnt, versionLn, medleyAppend, i: integer;
    sTrack, sArtist, sTitle, medleyTrack, prevMedTrack, sVersion, sSubVersion: string;

  begin
    iUpToRow := iRow + 1;
    if not on_FindRow( iUpToRow, 0, '<H2>Reviews</H2>', slMain ) then exit; // find last row with track info

    //the track super-loop.  the big while loop which looks for tracks.
    while iRow <= iUpToRow do begin
       
       //grab the track number off the current row.  current row will 
       //always contains a track number per while loop.
       iPos := Pos( 'e3b59d', slMain.Strings[ iRow ] );
       iPos2 := Pos( '.</TD>', slMain.Strings[ iRow ] );
       if (iPos > 0) And (iPos2 > 0) then begin
          sTrack := ( Trim( Copy( slMain.Strings[ iRow ], iPos+7, ipos2-(ipos+7) ) ) );
       end;
    
       //move forward 5 rows, and grab the track title
       if iRow <= iUpToRow then iRow := iRow + 5;
       sTitle := sys_RegexReplace( on_cleanHTMLLine( slMain.Strings[ iRow ] ), '[\(\):]', '', false )

//=============================================================================================
//======================= BEGIN version & sub-version string-grab BEGIN =======================
//=============================================================================================

   //enter the version/sub-version loop only if the verstion string for
   //the bootlegzone entry has data in it.  we can check this by looking for 
   //the presence of the default string, 'unidentified', in the version field.  
   //if we find that string, we'll skip grabbing version/sub-version strings

   //initiate strings to null
   sVersion := '';
   sSubVersion := '';

   versionLn := iRow;
   on_FindRow( versionLn, 0, 'version.php?vid=', slMain )

   if not sys_RegexFind ( slMain.Strings[ versionLn ], '.unidentified.') then
   begin
       //get the track version string 
       sVersion := on_cleanHTMLLine( slMain.Strings[ versionLn ] );

       //move forward two lines, and look for any sub-version string
       //if we find a the "version" URL, let's capture the sub-version string and 
       //compile the version and subversion strings into parenthetical information
       //which we'll append to the song title at the end
   
       if sys_RegexFind ( slMain.Strings[ versionLn + 2 ], '.version\.php.') then 
       begin
          sSubVersion := on_cleanHTMLLine( slMain.Strings[ versionLn + 2 ] );
          sVersion := '(' + sVersion + ', ' + sSubVersion + ')';
       end
       else
       begin
          sVersion := '(' + sVersion + ')';
       end;
 
       //Bootlegzone's default version nomenclature for "commercial release" versions
       //is the string 'Regular'.  I find this information useless, and therefore strip
       //any version information shown as 'Regular'.  Some users also enter multiple takes
       //under the 'Recording Sessions' version, with the take number as the sub-version.  
       //For my use, I would rather not have the 'Recording Sessions' nomenclature, so I kill it too

       //Note: I also kill some other common version tags... I need to upgade this list to a user-defineable array...
       if sys_RegexFind ( sVersion, '\(Regular\)') then sVersion := sys_RegexReplace( sVersion, '\(Regular\)', '', false);
       if sys_RegexFind ( sVersion, '\(Regular, ') then sVersion := sys_RegexReplace( sVersion, '\(Regular, ', '(', false);
       if sys_RegexFind ( sVersion, '\(Recording Sessions\)') then sVersion := sys_RegexReplace( sVersion, '\(Recording Sessions\)', '', false);
       if sys_RegexFind ( sVersion, '\(Recording Sessions, ') then sVersion := sys_RegexReplace( sVersion, '\(Recording Sessions, ', '(', false);
       if sys_RegexFind ( sVersion, '\(Recording Session\)') then sVersion := sys_RegexReplace( sVersion, '\(Recording Session\)', '', false);
       if sys_RegexFind ( sVersion, '\(Recording Session, ') then sVersion := sys_RegexReplace( sVersion, '\(Recording Session, ', '(', false);
       if sys_RegexFind ( sVersion, '\(Concert\)') then sVersion := sys_RegexReplace( sVersion, '\(Concert\)', '', false);
       if sys_RegexFind ( sVersion, '\(Concert, ') then sVersion := sys_RegexReplace( sVersion, '\(Concert, ', '(', false);
       if sys_RegexFind ( sVersion, '\(Radio\)') then sVersion := sys_RegexReplace( sVersion, '\(Radio\)', '', false);
       if sys_RegexFind ( sVersion, '\(Radio, ') then sVersion := sys_RegexReplace( sVersion, '\(Radio, ', '(', false);
       if sys_RegexFind ( sVersion, '\(Get Back Session\)') then sVersion := sys_RegexReplace( sVersion, '\(Get Back Session\)', '', false);
       if sys_RegexFind ( sVersion, '\(Get Back Session, ') then sVersion := sys_RegexReplace( sVersion, '\(Get Back Session, ', '(', false);
       if sys_RegexFind ( sVersion, '\(In Concert\)') then sVersion := sys_RegexReplace( sVersion, '\(In Concert\)', '', false);
       if sys_RegexFind ( sVersion, '\(In Concert, ') then sVersion := sys_RegexReplace( sVersion, '\(In Concert, ', '(', false);

   end;

//=============================================================================================
//=======================   END version & sub-version string-grab END   =======================
//=============================================================================================

//=============================================================================================
//=======================    BEGIN medley track naming routine BEGIN    =======================
//=============================================================================================

       //init variables needed for medley/subsong loop-check
       i := 1;
       medleyTrack := '';
       sameCnt := 1;

       //if a sub-song exists (checked on the page's html string 'TxNy' where x is the current sTrack
       //in the track super-loop and y is any number greater than zero), then loop through looking
       //for all subsongs of the main track (as in the case of a medley-type track), and concat them
       //into a single track title line, separated by pipes
       while on_FindRow( medleyLoopRow, 0, '<TR><A name=T' + sTrack + 'N' + InttoStr(i), slMain) do begin
         
          //clear the version stinge for medley tracks   
          sVersion := '';

          //if we're on the 1st medley track, make it so.  if we're on a subsequent
          //medley track, append that track to the existing string of medley tracks
          //with a pipe separator.  for a cleaner medley track title, if a medley
          //contains multiple versions of the same song in a row, the track title
          //entry will be shortened into "track title Nx" where N is the number of
          //times the same song appears back-to-back in the medley
          //
          //a limitation of this loop is that it's only coded to support medley tracks that 
          //contain between 0-99 versions of a song in the "Song Nx" format
          if (AnsiCompareText(medleyTrack, '') = 0) then 
          begin
             medleyTrack := Trim( on_cleanHTMLLine( slMain.Strings[ medleyLoopRow + 6 ] ) );
             if not sys_RegexFind( Copy( Trim(sTitle), Length(Trim(sTitle))-1, 9999), ':') then sTitle := sTitle + ':';

             //Some folks on bootlegzone start the medley track name with the 1st sub-track name,
             //this leads to a double-named medley entry string.  Hard to explain, but this if 
             //statement takes care of that.
             if AnsiCompareText(medleyTrack, Copy( sTitle, 1, Length(sTitle)-Length(medleyTrack))) <> 0 then sTitle := 'Medley:';  

             prevMedTrack := medleyTrack;
 
          end
          else
          begin          
             if (AnsiCompareText( prevMedTrack, Trim( on_cleanHTMLLine( slMain.Strings[ medleyLoopRow + 6 ] ) )) = 0) then 
             begin
                sameCnt := sameCnt + 1;
		
		//the medleyAppend integer holds the character length of a medley track's Nx appended string, 
		//this length is used several times through the loop to combine the medley sub-tracks.  The
		//value is the length sameCnt plus two chars for the whitespace and 'x'.	
                medleyAppend := Length(InttoStr(SameCnt))+2;

                //sys_MessageDlg('medleyAppend' + InttoStr(medleyAppend),3,0);
                //sys_MessageDlg('medleyTrack' + medleyTrack,3,0);

                //regex will match 0-199, but loop limited to supporting 0-99x repeat strings   
                if sys_RegexFind( Copy( medleyTrack, Length(medleyTrack)-medleyAppend, 9999), '[0-9]|[1-9][0-9]|1[0-9]x') then 
                begin
                   //a quick if-else that checks if we have a one-char or two-char sameCnt (<=10)
                   //so we can subtract the right amount of chars from the end of the track before
                   //continuing to combine it.  A one-char length sameCnt means we trim 3 chars 
                   //for 'x' and whitespace, a two-char sameCnt means trim 4.  
                   if (sameCnt <= 10) then medleyTrack := Copy( medleyTrack, 1, Length(medleyTrack)-3) + ' ' + InttoStr(sameCnt) + 'x'
                   else medleyTrack := Copy( medleyTrack, 1, Length(medleyTrack)-4) + ' ' + InttoStr(sameCnt) + 'x';
                   sTitle := Copy( sTitle, 1, Length(sTitle)-(Length(prevMedTrack)+medleyAppend));
                   prevMedTrack := Trim( on_cleanHTMLLine( slMain.Strings[ medleyLoopRow + 6 ] ) );
                   //sys_MessageDlg(medleyTrack,3,0); 
                end
                else
                begin
                   if sys_RegexFind( sTitle, '[0-9]|[1-9][0-9]|1[0-9]x') then sameCnt := 2;
                   medleyTrack := medleyTrack + ' ' + InttoStr(sameCnt) + 'x';
                   sTitle := Copy( sTitle, 1, Length(sTitle)-Length(prevMedTrack));
                   prevMedTrack := Trim( on_cleanHTMLLine( slMain.Strings[ medleyLoopRow + 6 ] ) );
                end;
             end
             else 
             begin
                medleyTrack := on_cleanHTMLLine( slMain.Strings[ medleyLoopRow + 6 ] );
                prevMedTrack := Trim( on_cleanHTMLLine( slMain.Strings[ medleyLoopRow + 6 ] ) );
             end;
          end;

          //lets append the medley tracks to the main track, but take care not to
          //append pipes for back-to-back medley tracks that were combined into Nx format
          //also watch out for main tracks with "ghost" medley tracks that got a colon
          //appended per the "1st medley track" if-statement above. 
          if sys_RegexFind( Copy( sTitle, Length(sTitle)-3, 9999), '\|') then sTitle := Copy( sTitle, 1, Length(sTitle)-3);
          if sys_RegexFind( Copy( Trim(sTitle), Length(Trim(sTitle))-1, 9999), ':') and (AnsiCompareText(medleyTrack, '') = 0) then sTitle := Copy( Trim(sTitle), 1, Length(Trim(sTitle))-1);
          if (AnsiCompareText(medleyTrack, '') <> 0) then sTitle := sTitle + ' | ' + medleyTrack;
                    
          medleyLoopRow := medleyLoopRow + 1;
          i := i + 1;
          end;

//=============================================================================================
//=======================      END medley track naming routine END      =======================
//=============================================================================================

//=============================================================================================
//=======================   BEGIN various artists naming routine BEGIN  =======================
//=============================================================================================

       //find the next lines which contain an artist name and track number
       //for "various artists" naming accounting
       nxtTrackLn := iRow;
       nxtArtistLn := iRow;
       on_FindRow( nxtTrackLn, 0, 'TD width=30 bgColor=#e3b59d', slMain )       
       on_FindRow( nxtArtistLn, 0, '<TD class=sa0>', slMain )

       //get string begin/end for extracting the track artist string from the found line
       iPos := Pos( '<TD class=sa0>', slMain.Strings[ nxtArtistLn ] );
       iPos2 := Pos( '</TD></TR></TBODY></TABLE>', slMain.Strings[ nxtArtistLn ] );

       //if you don't find an artist specified for the track, set the
       //track artist string to the album artist.
       //else set the artist to the string on the found line 
       if (nxtArtistLn > nxtTrackLn) And (iPos > 0) And (iPos2 > 0) then sArtist := on_getArtist
       else sArtist := ( Trim( Copy( slMain.Strings[ nxtArtistLn ], iPos+14, ipos2-(ipos+14) ) ) );

       //if the artist string for the track doesn't equal the artist for
       //the album, assume this is a various artists compilation and
       //default to the "artist_name - track_name" naming scheme
       if (AnsiCompareText(sArtist, on_getArtist) <> 0) then begin
          sTitle := sArtist + ' - ' + sTitle;
       end;

       //Doing last check of medley tracks to remove any colon/pipe combos and
       //pesky underscores that (honestly) result from lack of dutiful checking above.  
       //I look for malformed 'medley'/colon/pipe/underscore stuff and fix.
       sTitle := sys_RegexReplace( sTitle, '(:.\|)', ':', false);
       sTitle := sys_RegexReplace( sTitle, '(_)', '', false);
       sTitle := sys_RegexReplace( sTitle, 'Medley \|', 'Medley: ', false);

//=============================================================================================
//=======================     END various artists naming routine END    =======================
//=============================================================================================

//=============================================================================================
//=======================    BEGIN final track string cleanup BEGIN     =======================
//=============================================================================================

       //append the version info to the final track title
       sTitle := sTitle + ' ' + sVersion;
        
       //write the track
       on_addTrack( sTrack, sTitle, '', '' );

       //move the super-loop forward to the next title track
       while ( not on_FindRow( iRow, 0, 'TD width=30 bgColor=#e3b59d', slMain ) ) And ( iRow <= iUpToRow ) do begin
          iRow := iRow + 1;
       end;       
       
    end;

  end;

//=============================================================================================
//=======================      END final track string cleanup END       =======================
//=============================================================================================

//=============================================================================================
//=======================         END track title super-loop END        =======================
//=============================================================================================

  //---
  procedure GetLabel( sLine: string );
  var
    iPos, iPos2: integer;
  begin
    iPos := Pos( 'Publisher:', sLine );
    iPos2 := Pos( '<BR><B>Reference :', sLine );
    if iPos > 0 then begin
       on_setLabel( Trim( Copy( sLine, iPos+15, ipos2-(ipos+15) ) ) );
    end;
  end;


  //---
  procedure GetDate( sLine: string );
  var
    iPos, iPos2: integer;
  begin
    iPos := Pos( 'Date :', sLine );
    iPos2 := Pos( '<BR><B>Made', sLine );
    if iPos > 0 then begin
       on_setYear( Trim( Copy( sLine, iPos+10, ipos2-(ipos+10) ) ) );
    end;
  end;

  //---
  procedure GetComment( sLine: string );
  var
    iPos, iPos2: integer;
  begin
    iPos := Pos( 'Comments', sLine );
    iPos2 := Pos( '<FONT size=2>', sLine );
    if iPos > 0 then begin
       on_setComment( on_cleanHTMLLine (Trim( Copy( sLine, iPos+20, ipos2-(ipos+20) ) ) ) );
    end;
  end;

// -------------------- begin page parse --------------------

begin

  on_Init; // clear values

  slMain := TStringList.Create;
  try

    iRow := 0;
    on_loadHTML( slMain ); // load page to stringlist

    slMain.SaveToFile( 'c:\1.html' ); // save to a file to see what we have to deal with

    // artist - album
    if on_FindRow( iRow, 0, '<HEAD><TITLE>BootlegZone : ', slMain ) then GetArtistAlbum( slMain.Strings[ iRow ] );

    // album artwork
    if on_FindRow( iRow, 0, 'Scan submitted/created by', slMain ) then GetImage( slMain.Strings[ iRow ] );

    // date
    if on_FindRow( iRow, 0, 'BR><B>Date :', slMain ) then GetDate( slMain.Strings[ iRow ] );

    // label
    if on_FindRow( iRow, 0, '<TD align=left><B>Publisher:', slMain ) then GetLabel( slMain.Strings[ iRow ] );

    // comments
    if on_FindRow( iRow, 0, '</TABLE><BR><U><B>Comments', slMain ) then GetComment( slMain.Strings[ iRow ] );

    // tracks
    if on_FindRow( iRow, 0, 'TD width=30 bgColor=#e3b59d', slMain ) then GetTracks( slMain.Strings[ iRow ] );

  finally
    slMain.Free;
  end;

end.