'=================================================================
' LISTDUPE.BAS  by Brian McLaughlin, compile with PowerBASIC 3.x
'
' Searches C: drive for duplicate files and creates a file listing
' those duplicates, in the root dir of C: drive, named DUPEFILE.LST.
'=================================================================

$CPU 80386 'make it faster
$LIB ALL OFF 'make it smaller
$FLOAT PROCEDURE 'if you have a co-pro, use NPX instead

DECLARE SUB GatherNames( DirNames$(), File$(), FileDir%())
DECLARE SUB MakeDupeList( DirNames$(), File$(), FileDir%())
DECLARE SUB PrintTwirlBar( Counter% )

DIM DirName$( 1 TO 1400 ) 'handles up to 1400 directories
DIM FileName$( 1 TO 16384 ) 'handles up to 16K file names
DIM FileDir%( 1 TO 16384 ) 'indexes file names to dir names
SHARED gFileTotal% 'a global variable used in both SUBs

Synch! = TIMER: DO: Start! = TIMER: LOOP UNTIL Start! > Synch!

GatherNames DirName$(), FileName$(), FileDir%()
MakeDupeList DirName$(), FileName$(), FileDir%()

Elapsed! = TIMER - Start!: PRINT: PRINT "Elapsed time:"; Elapsed!

PRINT: PRINT "To find the file DUPEFILE.LST, look in root directory."
END

'=======================================================
SUB GatherNames( DirName$(), File$(), FileDir%()) STATIC
'=======================================================

  AllSpec$ = "*.*"
  Slash$ = "\"

  DirAttrib% = &H10
  AllAttrib% = &H10 OR &H2 OR &H4 'attrib for dirs, system and hidden files

  gFileTotal% = 0
  DirTotal% = 1
  DirIndex% = 1
  DirName$( 1 ) = "C:\"                        'to search current drive, just
                                               'use "\"

  PRINT "Collecting filenames... ";            'give 'em something to look at
  DO 'directory-switching loop
    ThisDir$ = DirName$( DirIndex% )           'search next dir
    PrintTwirlBar DirIndex%

    FileSpec$ = ThisDir$ + AllSpec$            'look for *.* in that dir
    ThisName$ = DIR$( FileSpec$, AllAttrib% )  'includes subdirs, hidden, syste

    IF LEN( ThisName$ ) THEN
      IF ( ATTRIB( ThisDir$ + ThisName$ ) AND DirAttrib% ) THEN
        INCR DirTotal%
        DirName$( DirTotal% ) = ThisDir$ + ThisName$ + Slash$
      ELSE
        INCR gFileTotal%
        File$( gFileTotal% ) = ThisName$
        FileDir%( gFileTotal% ) = DirIndex%
      END IF

      DO                       'main directory-searching loop
        ThisName$ = DIR$
        IF LEN( ThisName$ ) THEN
          IF ( ATTRIB( ThisDir$ + ThisName$ ) AND DirAttrib% ) THEN
            INCR DirTotal%
            DirName$( DirTotal% ) = ThisDir$ + ThisName$ + Slash$
          ELSE
            INCR gFileTotal%
            File$( gFileTotal% ) = ThisName$     'saves name of file
            FileDir%( gFileTotal% ) = DirIndex%  'index to pathname
          END IF
        ELSE
          EXIT DO              'if null name, then exit loop
        END IF
      LOOP
    END IF
    INCR DirIndex%
  LOOP UNTIL DirIndex% > DirTotal%

END SUB


'========================================================
SUB MakeDupeList( DirName$(), File$(), FileDir%()) STATIC
'========================================================

  CRLF$   = CHR$( 13, 10 )
  Indent$ = "  "
  Found$  = " was found in:" + CRLF$ + Indent$
  Header$ = "Total duplicate files found on C: drive:"
  Spacer$ = "      " 'saves room for number of dupes
  Accum$  = Header$ + Spacer$ + CRLF$ + CRLF$

  PRINT: PRINT "Sorting files..."
  ARRAY SORT File$( 1 ) FOR gFileTotal%, TAGARRAY FileDir%()

  OutFile$ = "C:\DUPEFILE.LST"
  BakFile$ = "C:\DUPEFILE.BAK"
  IF LEN( DIR$( BakFile$ ) ) THEN KILL BakFile$
  IF LEN( DIR$( OutFile$ ) ) THEN NAME OutFile$ AS BakFile$

  OPEN OutFile$ FOR BINARY AS #1

  Dupes% = 0
  ThisFile% = 1
  FileOne$ = File$( ThisFile% )

  PRINT "Collecting duplicates... ";
  DO
    LastNonDupe% = ThisFile%          'assume this file isn't a dupe
    DO
      INCR ThisFile%                  'look at the next adjacent file
      PrintTwirlBar ThisFile%         'give the user something to stare at
      FileTwo$ = File$( ThisFile% )
      IF LEN( FileOne$ ) = LEN( FileTwo$ ) THEN 'eliminates some files quickly
        IF FileOne$ = FileTwo$ THEN   'if equal, we found a dupe

          LastDupe% = ThisFile%
          ThisDirNum% = FileDir%( ThisFile% )
          Dir2$ = DirName$( ThisDirNum% ) + CRLF$

          IF LastNonDupe% = ( ThisFile% - 1 ) THEN
            PrevDirNum% = FileDir%( LastNonDupe% )
            Dir1$ = DirName$( PrevDirNum% ) + CRLF$
            Accum$ = Accum$ + FileOne$ + Found$ + Dir1$ + Indent$ + Dir2$
            INCR Dupes%, 2            'count both this and previous file
          ELSE
            Accum$ = Accum$ + Indent$ + Dir2$
            INCR Dupes%
          END IF
          IF LEN( Accum$ ) > 16384 THEN  'needed to help avoid running
            PUT$ #1, Accum$              'out of RAM if many dupes found
            Accum$ = ""
          END IF
        ELSE
          EXIT DO
        END IF
      ELSE
        EXIT DO
      END IF
    LOOP

    IF ThisFile% = gFileTotal% THEN EXIT DO   'we're done - let's split!

    IF LastDupe% = ( ThisFile% - 1 ) THEN     'this dupe group complete
      Accum$ = Accum$ + CRLF$                 'add a blank line after
    END IF

    SWAP FileOne$, FileTwo$    'a wee bit faster than:  Two$ = One$
  LOOP

  TotalDupes$ = STR$( Dupes% )

  PUT$ #1, Accum$              'write the dupe list
  SEEK #1, LEN( Header$ )      'go back and fill in
  PUT$ #1, TotalDupes$         'the number of duplicate files
  CLOSE #1

END SUB

'===============================
SUB PrintTwirlBar( Counter% )
'===============================

  LOCATE CSRLIN,( POS )
  PRINT MID$( "/-",( Counter% MOD 4 ) + 1, 1 );

END SUB