PACKAGE |STAT Data Manipulation and Analysis, by Gary Perlman
NAME dsort - multiple key data-file sorting filter
SYNOPSIS dsort [-aceinr] [-l lines] [[type][column-range]] ...
DESCRIPTION dsort sorts data it reads from the standard input. It sorts lines based on numerical or alphabetical values in whitespace-separated columns. For when two columns have the same value, you can supply further sorting columns to break ties. dsort sorts by looking at columns you choose, and orders lines so that the data or labels in the chosen columns are ordered. You can name the specific column numbers you want, and you can use the M-N notation to specify a sequence, from M to N, of columns. Ranges can be from small to large or from large to small column numbers. If you don't supply columns to sort with, dsort uses column 1, then 2, then 3, and so on, until it finds a difference, or all columns are used.

You can supply the method for comparing column values: a, for alphabetical, i, for integer, n, for most other numbers, including those with decimal points, or e, for sorting all types of numbers, even ones with exponential scientific notation. If you do not supply the type of sort, dsort chooses the one that is best suited for the data, although this makes dsort do some extra work. For alphabetical sorts, you can request a case-insensitive sort with c; that would make hello match Hello and HELLO. You can reverse the sort from the default ascending order to descending order with r. These sorting types can be specified globally for all columns at once with command line options. Global types can be over-ridden by supplying options for specific columns or ranges of columns.

OPTIONS
-a
Sort fields by alphabetic ordering of columns.
-c
Use a case-insensitive comparison for alphabetic comparisons. When used as a type modifier for a specific range of columns, this also forces alphabetic sorting. As a command line option, it only modifies the sorting of alphabetic columns.
-e
Sort fields by exponential notation ordering of columns. This is the only true numeric sort that works for all numbers: integers, numbers with decimal points, and exponential (scientific) notation numbers. But it is a lot slower than the other sorting options, especially on machines without floating point hardware.
-i
Sort fields by integer ordering of columns.
-l lines
Set the maximum number of lines to be read.
-n
Sort fields by special numeric ordering of columns. Numbers with decimal points are allowed, and compared by a special fast comparison method that is almost as fast as integer sorting.
-r
Reverse the order of comparison for sorting options.
The following standard help options are supported. The program exits after displaying the help.
-L
Display limits
-O
Display options and values
-V
Display version number and date
EXAMPLES Suppose the file ex.dat has the contents below. If you did not specify column types, dsort would infer that columns 1 and 2 were to be sorted alphabetically, column 3, as integers, and column 4, numerically.

# file with four columns (alpha, alpha, integer, numerical) 

cat ex.dat
red       high      1         3.14
blue      low       2         1.62
green     high      6         2.54
green     low       4         2.71
blue      high      3         1
red       low       5         1
# sort by column 1, then 2, then 3, then 4 
dsort  <  ex.dat   
blue      high      3         1
blue      low       2         1.62
green     high      6         2.54
green     low       4         2.71
red       high      1         3.14
red       low       5         1
# sort column 3 in ascending order 
dsort  i3  <  ex.dat   
red       high      1         3.14
blue      low       2         1.62
blue      high      3         1
green     low       4         2.71
red       low       5         1
green     high      6         2.54
# numerically sort column 4 in reverse order, then 3 
dsort  rn4  3  <  ex.dat   
red       high      1         3.14
green     low       4         2.71
green     high      6         2.54
blue      low       2         1.62
blue      high      3         1
red       low       5         1
# reverse numerical sort of column 4 and then 3 
dsort  rn4  r3  <  ex.dat   
red       high      1         3.14
green     low       4         2.71
green     high      6         2.54
blue      low       2         1.62
red       low       5         1
blue      high      3         1
# sort by column 2, then column 1 within ties in 2 
dsort  2-1  <  ex.dat   
blue      high      3         1
green     high      6         2.54
red       high      1         3.14
blue      low       2         1.62
green     low       4         2.71
red       low       5         1
LIMITS Use the -L option to determine the program limits.
SEE ALSO maketrix to format matrix format files.
transpose to transpose matrix format files.
colex to extract columns.
perm to randomize or sort line-oriented files.
UPDATED June 17, 1986