readme.txt

能够帮您找出来在硬盘中重复出现的文件

                Find Duplicates - discover duplicated files

Find Duplicates was written to allow you to control your disk space usage by 
discovering files that are duplicated and, should you so wish, deleting one or 
more of these duplicates.  There are many ways in which duplicate files can be 
deposited on your hard disk, for example programs which don't check to see if 
you have a particular DLL installed and install their own private copy in any 
case, or other programs that install a DLL in your \Windows folder when it is 
already in \Windows\System32.  You can also use Find Duplicates to see if any 
files on a floppy are already present anywhere on your hard disk.  For safety,
Find Duplicates moves files to the recycle bin before you delete them.


How does Find Duplicates work?

Find Duplicates scans one or more disks on your system to find multiple files, 
in a two-phase process.  First it scans all the folders and sorts all the files 
it finds into size order (files HAVE to be the same size to be identical - 
yes?)  You can limit the scan to one folder tree, if you wish.  It then compares 
files of the same size to see if the contents are actually identical, and lists 
identical files by size order.  You can then double-click on any file to examine 
its properties, and optionally move it to the recycle bin.

This process can take some time, so Find Duplicates will first perform one of 
two preliminary checks to see if the files might actually be identical without 
having to actually examine the whole file.  By default, it checks the 
modification date and time of the files, and only compares the files 
byte-by-byte if the timestamps are the same.  But it is possible for two files 
to have the same contents without having the same timestamp, so you can enable 
an option whereby the first 512 bytes of each file are checksummed.  This 
improves the recognition of identical files, but it is slower, and since it 
involves a file access, the file's last access date will be altered.  

By default, the timestamp, not the checksum comparison is selected.  In either 
case, the filename is normally ignored, so simply renaming a file will not hide 
the fact that it is a duplicate.  The timestamp of zero size files is ignored.  
If you wish, you can also require that duplicate files must have the same file 
name.  You may be rather surprised to discover what duplicates by content 
actually exist in some popular office suites!


When should I disable timestamp checking?

You can turn off the timestamp checking in favour of the slower checksum method 
should you so wish.  For example, if many identically sized and timestamped 
files are found from the initial search, using just timestamps might miss some 
duplicate files since the duplicates may not be adjacent in the name ordered 
list produced by the folder scan.  The program was not designed for this sort of 
duplicate search, but will perform adequately with timestamp checking turned 
off.  You might also wish to disable timestamp checking if you suspected that 
different products had installed identical support DLLs.


Usage:

Extract FindDupl.exe from the zip file to a convenient location, and run it!  
Only the FindDupl.exe file is required from the archive.  You will be presented 
with a dialog box showing you disk drives, with your local hard disk drives 
selected.  You can optionally enter a file spec such as *.EXE and a folder 
specification such as \windows to limit the search.  Note that if you enter a 
folder specification, only that folder will be searched on each drive (e.g. 
c:\windows, d:\windows and so on).  Press the Start Search button to find 
duplicate files.

If you don't already have Borland's run-time library, you will also need to 
download the file VCL30.zip from any site offering it.  For example:
  http://www.user.globalnet.co.uk/~hermes02/Freeware/VCL30.zip
Extract the file VCL30.DPL from this archive and copy to your Windows System
folder (e.g. \Windows\System or \WinNT\System32).  

There is a status bar which will keep you informed on the progress of both the 
folder scan phase, and the file comparison phase.  Once the main list box has 
filled up with file names, you can double-click on a file name to get a pseudo 
Properties dialog box (actually written in Delphi, not derived from the system 
right-click -> Properties box).  You will see a delete button which allows you 
actually to delete the file.

If a floppy disk (specifically drive A:) is included in the selected drives to 
scan, the program will normally assume that you wish to find files in common 
between the floppy and the other disk drives, so that during the folder scan 
phase Find Duplicates will only record files on the other drives that are the 
same size as files found on the floppy.  This makes the scanning faster and 
allows you to ask the question "Do I already have any files on my hard disk that 
are on this floppy?"  You can treat floppies just as ordinary disks by 
unchecking the "Treat floppy as master" check box.  You may notice a slightly 
different message in the status bar during the folder scan phase in this case.

Windows 95 has a special hidden folder called SYSBCKUP where backup copies of 
critical system files are stored.  Find Duplicates will recognise a folder with 
\SYSBCKUP\ in the path name, and ignore any files in that folder.  To disable 
this safety feature, uncheck the "Skip SYSBCKUP folder" check box.  The status 
bar will indicate that the folder is being skipped, but you'll have to be quick 
to see that message!  Other hidden folders are scanned normally, except that
folders which have the file DESKTOP.INI, and are therefore special folders, are
skipped (e.g. fonts, Internet Explorer history and channels).

Multiple folder trees to scan can be specified in the "Look in:" edit box, with 
the folder names separated by commas.  Folder names containing spaces must be 
entered with quotes - "\My Files".  A leading "\" will be supplied if you omit 
it.  If you enter multiple folders, and one folder contains another the 
contained folder is ignored so that the contained folder is not scanned twice.  
For example, if \Windows and \Windows\System had been entered, \Windows\System
would be removed from the list of folders to scan as files there would be found 
when the \Windows tree was scanned.

Should you wish to skip one or more folders, you can enter a single folder or a
list of folders to skip.  Suppose you have files from a service pack that you 
know are duplicated in your Windows directory, and that these files are stored
in a folder called SP3.  Entering SP3 as the folder to skip will cause all 
folders having \SP3\ as part of the path name to be ignored.  You can use comma 
separated text to enter more than on folder - SP2, SP3.  If your folder 
contains spaces, you'll need to enclose the name in quotes - "Program Files".
Simply entering "\" will be ignored.  The list of folders to skip will be saved 
in the registry between runs.

Find Duplicates will ignore files that have zero length, because the data in 
such files does not occupy disk space, and they are often simply marker files 
(e.g. hidden files to show that a folder was created by installing an 
application and not a user).  If you prefer to find these files, uncheck the 
"Skip zero-length files" check box.  Be aware that these files actually take up 
at least 32 bytes of directory space, but that since the folder must be at least 
a cluster size long (e.g. 4096, 8192 bytes) there will typically be very little 
overhead for a single zero-length file within a moderately full folder.

From version 1.4 onwards, the program will allow you to find "duplicates" which
are not the same size.  This special mode will help those trying to sort out
multiple incompatible DLLs (such as CTL3D.dll) which can get dumped on your
system.  In this mode, internally, all files are 1234567890 bytes long.  I'd 
welcome comments on the usefulness and implementation of this mode.

Upon exiting, Find Duplicates will try to save the list of duplicates in a file 
named FindDupl.lis in the same folder as the FindDupl.exe program file.  If this 
file is present on starting the program, Find Duplicates will ask if you would 
like to reload the list.  This allows you to split the task of deleting of 
duplicate files into short sessions without having to run the time consuming 
scan and compare phases every time.  Do not edit FindDupl.lis.

Find Duplicates will save a list of duplicates it has discovered in NewDupl.lis
in the same folder as the program.  You may copy this list to a file named
SkipDupl.lis which may be used to skip duplicates that are already known.  The
two separate files allow you to build up a composite list for your whole system
from a number of separate runs, adding in parts of NewDupl.lis to SkipDupl.lis.
Note that you should only list the duplicate files, not the master files, in
SkipDupl.lis, otherwise when new duplicate files appear on the disk, there will
be nothing for them to match against, and they will not be detected.
Eventually, I may make this a drag-and-drop function within the program.

For safety, Find Duplicates will not actually delete files, but instead will 
move them to the Recycle Bin.  This means that the disk space will not actually 
be returned until the Recycle Bin is emptied.  Right-click on the Recycle Bin to 
access the Empty Recycle Bin function.


  +------------------------------ WARNING ---------------------------------+
  |                                                                        |
  |  You take sole responsibility if you choose to delete a file.  Find    |
  |  Duplicates makes no attempt to check if the file is in use or key to  |
  |  the functioning of your computer. Take backups before making changes. |
  |                                                                        |
  +------------------------------ WARNING ---------------------------------+


Notes:

The program is written with Borland's Delphi 3.02, and most of the source code 
is included.  You do not need access to Delphi to run Find Duplicates.  You will 
need other Delphi units (not included in the .ZIP file) in order to recompile 
Find Duplicates.  The program requires Windows 95 or NT 4.0.

The folder scan phase can consume a large amount of virtual memory if a wildcard 
*.* is specified.  At present, the program does not detect when its memory 
allocations fail, and may hang in these circumstances with an out-of-memory 
error.  Increase the space available for the Windows swapfile or avoid 
specifying wildcards if this happens to you.  This seems to have been resolved
somewhat with version 3.02 of Delphi.


Release information:

1997 Jan 29  V1.0.4  First released version
1997 Feb 03  V1.0.6  Treat floppy drive as master
1997 Feb 12  V1.0.8  Decode date of "0" as "unknown" on Properties dialog
1997 Apr 02  V1.1.0  Make file list box hint the filename (for long paths!)
                     Save and optionally restore duplicate file list
                     By default, ignore files in Win 95 SYSBCKUP folder
                     Replace ListBox with ListView (both Drives and Results)
                     Correct: missing FindClose in do_checksum routine
                     Correct: remove deleted file from the duplicates list
1997 Apr 07  V1.1.2  Use ShellAPI function to move file to recycle bin
1997 May 13  V1.1.4  Use my own TFileList component
                     Don't show properties/delete box for non-existant files
                     Put source files in sub-folder
                     Force checksum routine to return 31-bit value 
1997 May 18  V1.2.0  Move to Delphi 3.0
                     Don't leave singletons in the duplicates list
                     Correct property display for sequential compressed files
                     Don't allow ColumnClick on the FileListView - set False
1997 Oct 08  V1.2.2  Move to Delphi 3.01
                     Handle large font displays better.
                     Use TreeScanner with FindHiddenXX options
                     Don't build against run-time VCL30.DPL
1997 Nov 24  V1.2.4  Move to Delphi 3.02
                     Do build against run-time VCL30.DPL
1997 Nov 29  V1.2.6  Allow a list of folders to ignore
1998 Jan 24  V1.2.8  Save and restore ignore folder list using registry
                     Save and restore edit and check box settings
                     Save a list of found duplicate file names as "NewDupl.lis"
                     Optionally read "SkipDupl.lis" an ignore list on next scan
                       (pending drag-and-drop generation of this list)
                     Skip special folders that have DESKTOP.INI present
                     Add "File names must be same" option (disabled by default)
1998 Feb 23  V1.3.0  Add file version column
1998 Feb 28  V1.3.2  Allow for scanning multiple root folders
                     Handle zero matching files found correctly (oops!)
                     Add scan details to program caption
1998 Mar 15  V1.4.0  Allow finding multiple file masks
                     Allow for ignoring file size (i.e show all DLL versions)
                     Add auto-size of displayed column widths
1998 Mar 16  V1.4.2  Correct problems in FileListView when columns missing
                     Sort "find-all" (ignore size) output into file name order
                     Add program version information

Contacting the author:

This program is freeware, and remains copyright of David J Taylor, Edinburgh, 
1997-8.  This program is provided "as is", without any support.  Whilst I cannot 
answer queries relating to the use of this program, I'd welcome any comments or 
suggestions for improvements you may have, and I would like to thank those who
have contributed such feedback which has helped mould the present version of 
the program.


[e-mail]
1998 March 16