msort(USP)

Name

msort : In-Memory Sort

Synopsis

Usage   : msort key=<key> <file>

Options : -e

          -s<c>

          -p<n>

          --stable

Version : Tue Jan  9 09:02:34 JST 2024

Edition : 5

Description

Sorts a file on specfied key by <key>.  <key> designates the field

position as fllows:

single field       2        the 2nd field

                   NF   the last field

                   NF-1   the field just before the last field

contiguous fields  2/4      from the 2nd field to the 4th field

                   4/2   from the 4th field to the 2nd field

                   NF-3/NF  from NF-3 field to the NF field

combination        2@NF     the 2nd field and the NF field

There is no limit on the length of the key field or on the number

of key fields.  The key field can also contain multi-byte characters

such as Japanese.

If you specify "r" as sort method after the field position, the

values will be sorted in descending order.  If you specify "n" as

sort method after the field position, that field's values will be

sorted as numbers.  If you specify "nr" as sort method after the

field, the values will be sorted in descending order as numbers.  If

you specify sort method before or after the "/", you must use the

same sort method for both fields.

  2n/5n     OK

  2n/5nr    Error

  2n/5r     Error

When you specify "e" as sort method or specify -e ootion and no

method, characters in the field are replaced as follows and compared

as string:

  _  ==> 0x20 (space)

  \0 ==> 0x00 (null)

  \t ==> 0x09 (tab stop)

  \n ==> 0x0a (new line)

  \r ==> 0x0d (carrige return)

  \_ ==> 0x5f (underscore)

  \\ ==> 0x5c (back slash)

If the file name is omitted or if it is specified as "-" then the

command will read from standard input.

If you specify the -p<n> option, the command will run in <n>

parallel processes.  There is no limit to the number of parallel

processes you can specify, but if you specify more processes

than there are cores available on your CPU then performance

will be degraded.  If you are sorting less than 10,000 lines,

even if you specify the -p<n> option it will only run in single

process.

--stable is prepared for compatibility with isort, and ignored by

this command.

Important

msort uses the merge sort algorithm, so there is no danger that the

processing time will be extremely long no matter what state the

input data is in.

Note

msort is an in-memory sort, so it will use at a maximum three times

the memory as the size of the file being sorted.