.\"
.\" To format this file into a text file say
.\"
.\"    nroff -man samefile.1              
.\"
.\" or
.\"
.\"    groff -man -Tascii samefile.1
.\"
.TH SAMEFILE 1 "OCTOBER 1997" IOCCC "User Manuals"
.SH NAME
samefile \- find identical files
.SH SYNOPSIS
.B samefile
.SH DESCRIPTION
.B samefile
reads a list of file names (one file name per line) from stdin.
For each file name pair with identical contents, a
line consisting of six tab separated fields is output:
The size in bytes, two file names, the character ``='' if
the two files are on the same device, ``X'' otherwise,
and the link counts of the two files.
The output is sorted in reverse order by size.

.B samefile
uses two stages to give optimum performance.
 
In the first stage, all non-plain files are silently
ignored (directories, devices, FIFOs, sockets,
symbolic links) as well as files for which
.BR stat (2)
or
.BR open (2)
fails.
The result of the first stage (the file names) is written into
a binary tree with one node for every file size. Each node
is in turn a linked list of file names along with inode and
device information.
It is also at this first stage where checks for hard links are done.
For any inode only one filename will be added to the binary tree.

In the second stage all files having the same
size are compared against each other. The rules
of mathematical logic are applied to reduce work
and output noise: if files A, B, and C have the same size and
.B samefile
finds that A = B and A = C then it will not
compare B against C (and will not
output a line for B and C) but only
for A = B and A = C. The algorithm will detect
equality across arbitrarily long chains.
Note however, that because
only the first filename per inode gets into the
first stage, the output for a group of identical files
with different inode numbers is also minimized. Suppose
you have six identical files of size 100
in an inode group consisting of the three
inodes with numbers 10, 20 and 30:
 
.nf
$ ls -i   # output edited for readabilty:
   10 file1     20 file4     30 file6
   10 file2     20 file5
   10 file3
$ ls | samefile
100     file1   file4   =       3       2
100     file1   file6   =       3       1
 
.fi
The sum of the sizes in the first column is the amount of disk space
you could gain by making all 6 files links to only one
file or remove all but one of the files. To be precise,
disk space is allocated in blocks - you will probably gain
two blocks here, rather than 200 bytes.
Note that it is not enough to just remove file4 and
file6 (you would gain only 100 bytes because file5
still exists.)

.SH LIMITATIONS
.B Samefile
was written with no limits in mind. The number of input lines is
unlimited. The size of the actual files is only limited by
available virtual memory needed to compare one pair of files.
The only hard limit is the constraint that there should not
be more than about 8192 files having the same size. Experience
has shown that there are rarely more than a couple dozen
files of the same size.
.SH EXAMPLES
.nf

For everybody:
What are the duplicates under my home directory?

    $ find $HOME | samefile

For the sysadmin folks:
Report all duplicate files under /usr larger than 16k:

    $ find /usr -size +16384c -a -type f | samefile

For the ftp and WWW admins:
How much space is wasted below our site's /pub directory?

    $ find /pub -type f | samefile | awk '{sum += $1} END {print sum}'

.fi

.SH "EXIT STATUS"
If
.B samefile
runs out of memory the exit status is 1,
zero otherwise.
.fi
.SH "SEE ALSO"
.BR find (1),
.BR ln (1),
.BR rm (1)
.\"
.\"            End Of Man Page