Find Duplicate Files (based on size first, then MD5 hash)

find -not -empty -type f -printf "%s\n" | sort -rn | uniq -d | xargs -I{} -n1 find -type f -size {}c -print0 | xargs -0 md5sum | sort | uniq -w32 --all-repeated=separate
This dup finder saves time by comparing size first, then md5sum, it doesn't delete anything, just lists them.

73
By: syssyphus
2009-09-21 00:24:14

9 Alternatives + Submit Alt

  • If you have the fdupes command, you'll save a lot of typing. It can do recursive searches (-r,-R) and it allows you to interactively select which of the duplicate files found you wish to keep or delete.


    22
    fdupes -r .
    Vilemirth · 2011-02-19 17:02:30 0
  • Calculates md5 sum of files. sort (required for uniq to work). uniq based on only the hash. use cut ro remove the hash from the result.


    18
    find -type f -exec md5sum '{}' ';' | sort | uniq --all-repeated=separate -w 33 | cut -c 35-
    infinull · 2009-08-04 07:05:12 1
  • Improvement of the command "Find Duplicate Files (based on size first, then MD5 hash)" when searching for duplicate files in a directory containing a subversion working copy. This way the (multiple dupicates) in the meta-information directories are ignored. Can easily be adopted for other VCS as well. For CVS i.e. change ".svn" into ".csv": find -type d -name ".csv" -prune -o -not -empty -type f -printf "%s\n" | sort -rn | uniq -d | xargs -I{} -n1 find -type d -name ".csv" -prune -o -type f -size {}c -print0 | xargs -0 md5sum | sort | uniq -w32 --all-repeated=separate Show Sample Output


    2
    find -type d -name ".svn" -prune -o -not -empty -type f -printf "%s\n" | sort -rn | uniq -d | xargs -I{} -n1 find -type d -name ".svn" -prune -o -type f -size {}c -print0 | xargs -0 md5sum | sort | uniq -w32 --all-repeated=separate
    2chg · 2010-01-28 09:45:29 0
  • This works on Mac OS X using the `md5` command instead of `md5sum`, which works similarly, but has a different output format. Note that this only prints the name of the duplicates, not the original file. This is handy because you can add `| xargs rm` to the end of the command to delete all the duplicates while leaving the original.


    2
    find . -type f -exec md5 '{}' ';' | sort | uniq -f 3 -d | sed -e "s/.*(\(.*\)).*/\1/"
    noahspurrier · 2012-01-14 08:54:12 4
  • Finds duplicates based on MD5 sum. Compares only files with the same size. Performance improvements on: find -not -empty -type f -printf "%s\n" | sort -rn | uniq -d | xargs -I{} -n1 find -type f -size {}c -print0 | xargs -0 md5sum | sort | uniq -w32 --all-repeated=separate The new version takes around 3 seconds where the old version took around 17 minutes. The bottle neck in the old command was the second find. It searches for the files with the specified file size. The new version keeps the file path and size from the beginning.


    1
    find -not -empty -type f -printf "%-30s'\t\"%h/%f\"\n" | sort -rn -t$'\t' | uniq -w30 -D | cut -f 2 -d $'\t' | xargs md5sum | sort | uniq -w32 --all-repeated=separate
    fobos3 · 2014-10-19 02:00:55 0

What Others Think

As an alternative, check out http://www.pixelbeat.org/fslint/ in case you don't mind using a GUI for this. It gives you the option of hard linking the duplicate files and doing other lint-y tasks. Available as package 'fslint' at least in debian/ubuntu.
bwoodacre · 456 weeks and 4 days ago
Thanks for the FSlint reference. Note fslint uses much the same mechanism underneath and has a CLI mode. http://fslint.googlecode.com/svn/trunk/fslint/findup
pixelbeat · 456 weeks and 4 days ago
awsome, much faster then fdupes.
dakunesu · 456 weeks and 3 days ago
Isn't the -D redundant?
dennisw · 455 weeks ago
yes it is... thanks for noticing, I fixed it.
syssyphus · 455 weeks ago
How can you mass delete these files once they're found? (I'd like to keep one of them)
matthewbauer · 451 weeks ago
you might want to look at fdupes or fslint in order to help with hardlinking / deleting, etc... my command is really just a quick hack to list them.
syssyphus · 451 weeks ago
There is also perfect match: http://pmatch.rubyforge.org/ That's especially if you are commandline fan.
zabuch · 438 weeks and 1 day ago
Fantastic, man. this is truly great.
oernii3 · 379 weeks and 5 days ago
There is also rmlint: https://github.com/sahib/rmlint Example: rmlint [path] -GYX -v5 + Gives you similiar results + you can pipe it directly to 'sh' + it's lots faster as additionally fingerprints are done and a few other tricks. + it has also other options ;-)
sahib · 377 weeks and 5 days ago
"find -type" doesn?t work on Mac OS X.
ELV1S · 318 weeks and 4 days ago
can filename comparison be added as a first step to the first solution given? find -not -empty -type f -printf "%s\n" | sort -rn | uniq -d | xargs -I{} -n1 find -type f -size {}c -print0 | xargs -0 md5sum | sort | uniq -w32 --all-repeated=separate seems to me checking filename first could speed things up. if two files lack the same filename then in many cases i would not consider them a dupe. thanks
johnywhy · 310 weeks and 4 days ago
The code for findup by Pádraig Brady (http://fslint.googlecodesh.com/svn/trunk/fslint/findup) is very OS (or user defined system) sensitive and is without comments that tell you what it is pointing to: --------- ./FindDups: line 62: /Programming/FSlint/supprt/fslver: No such file or directory ./FindDups: line 135: shell_quote: command not found ./FindDups: line 147: /Programming/FSlint/supprt/getfpf: No such file or directory ./FindDups: line 149: check_uniq: command not found ./FindDups: line 164: /Programming/FSlint/supprt/rmlint/merge_hardlinks: No such file or directory --------- /Programming is my partition for assorted programming I am doing. I use openSUSE 12.1. I would assume (with all that connotes) that uniq could be used rather than check_uniq, and that the . /supprt directory is unique to another distro (why is there so much illogical difference - that eliminates a lot of people who would like to switch from Windows). Either that or it is one of Pádraig Brady's personnal directories and that does not fly unless they are included. Considering this came from Google code, you have to first assume it is incomplete. And this is no exception to that!
JohnLB · 308 weeks and 2 days ago
Hi, I would like if it is any way of find duplicates of a given file (not all duplicates on the fs) maybe searching directly by md5. It would be great for me.
tia · 271 weeks and 3 days ago
Does anyone know a way to find duplicate files between 2 volumes based only on size and name?
d0g · 228 weeks and 5 days ago
kos omkon 3a hal exemple
mahmoud · 211 weeks and 1 day ago
Here is the tool to find and delete duplicate files "DuplicateFilesDeleter"
ivanden · 205 weeks and 1 day ago
Should be noted that md5sum is not a collision free algorithm, so there's a probability (OK, a very small probability) that the above commands will reports files as dupes even when they're not. Should use a sha1sum if you're paranoid.
befyber · 181 weeks and 4 days ago
Just a thot: Sequence of detection 1. size 2. MAX=say 100Mb or 10 Mb (a) Full MD5 for files (b) MD5 only the first MAX bytes for files > MAX e.g. MD5 (dd if=file of=/tmp/file count=200000) 3. Full MD5 or SHA1 for files found in 2(b) --- Wud b nice for media collection. Any takers??
Atanu · 176 weeks and 3 days ago
Duplicatefilesdeleter is best removal for duplicate files
Ketan · 150 weeks and 2 days ago
I use Duplicate Files Deleter as it is very effective. It is 100% accurate and performs the scan quickly.
rewanya · 127 weeks and 2 days ago
Please use Duplicate files deleter, it is very simple to use. But make it sure to keep the important files in backup. Thanks everybody for giving me your valuable times.
TinaRodrigo · 95 weeks and 3 days ago
ffgg
chenlixiang · 46 weeks and 1 day ago

What do you think?

Any thoughts on this command? Does it work on your machine? Can you do the same thing with only 14 characters?

You must be signed in to comment.

What's this?

commandlinefu.com is the place to record those command-line gems that you return to again and again. That way others can gain from your CLI wisdom and you from theirs too. All commands can be commented on, discussed and voted up or down.

Share Your Commands



Stay in the loop…

Follow the Tweets.

Every new command is wrapped in a tweet and posted to Twitter. Following the stream is a great way of staying abreast of the latest commands. For the more discerning, there are Twitter accounts for commands that get a minimum of 3 and 10 votes - that way only the great commands get tweeted.

» http://twitter.com/commandlinefu
» http://twitter.com/commandlinefu3
» http://twitter.com/commandlinefu10

Subscribe to the feeds.

Use your favourite RSS aggregator to stay in touch with the latest commands. There are feeds mirroring the 3 Twitter streams as well as for virtually every other subset (users, tags, functions,…):

Subscribe to the feed for: