OpenBSD manual page server

Manual Page Search Parameters

SORT(1) General Commands Manual SORT(1)

sortsort, merge, or sequence check text and binary files

sort [-bCcdfgHhiMmnRrsuVz] [-k field1[,field2]] [-o output] [-S size] [-T dir] [-t char] [file ...]

The sort utility sorts text and binary files by lines. A line is a record separated from the subsequent record by a newline (default) or NUL ‘\0’ character (-z option). A record can contain any printable or unprintable characters. Comparisons are based on one or more sort keys extracted from each line of input, and are performed lexicographically, according to the specified command-line options that can tune the actual sorting behavior. By default, if keys are not given, sort uses entire lines for comparison.

If no file is specified, or if file is ‘-’, the standard input is used.

The options are as follows:

, --check=silent|quiet
Check that the single input file is sorted. If it is, exit 0; if it's not, exit 1. In either case, produce no output.
, --check
Like -C, but additionally write a message to if the input file is not sorted.
, --merge
Merge only; the input files are assumed to be pre-sorted. If they are not sorted, the output order is undefined.
output, --output=output
Write the output to the output file instead of the standard output. This file can be the same as one of the input files.
size, --buffer-size=size
Use a memory buffer no larger than size. The modifiers %, b, K, M, G, T, P, E, Z, and Y can be used. If no memory limit is specified, sort may use up to about 90% of available memory. If the input is too big to fit into the memory buffer, temporary files are used.
Stable sort; maintains the original record order of records that have an equal key. This is a non-standard feature, but it is widely accepted and used.
dir, --temporary-directory=dir
Store temporary files in the directory dir. The default path is the value of the environment variable TMPDIR or /tmp if TMPDIR is not defined.
, --unique
Unique: suppress all but one in each set of lines having equal keys. This option implies a stable sort (see below). If used with -C or -c, sort also checks that there are no lines with duplicate keys.

The following options override the default ordering rules. If ordering options appear before the first -k option, they apply globally to all sort keys. When attached to a specific key (see -k), the ordering options override all global ordering options for that key. Note that the ordering options intended to apply globally should not appear after -k or results may be unexpected.

, --dictionary-order
Consider only blank spaces and alphanumeric characters in comparisons.
, --ignore-case
Consider all lowercase characters that have uppercase equivalents to be the same for purposes of comparison.
, --general-numeric-sort, --sort=general-numeric
Sort by general numerical value. As opposed to -n, this option handles general floating points. It has a more permissive format than that allowed by -n but it has a significant performance drawback.
, --human-numeric-sort, --sort=human-numeric
Sort by numerical value, but take into account the SI suffix, if present. Sorts first by numeric sign (negative, zero, or positive); then by SI suffix (either empty, or `k' or `K', or one of `MGTPEZY', in that order); and finally by numeric value. The SI suffix must immediately follow the number. For example, '12345K' sorts before '1M', because M is "larger" than K. This sort option is useful for sorting the output of a single invocation of 'df' command with -h or -H options (human-readable).
, --ignore-nonprinting
Ignore all non-printable characters.
, --month-sort, --sort=month
Sort by month abbreviations. Unknown strings are considered smaller than valid month names.
, --numeric-sort, --sort=numeric
An initial numeric string, consisting of optional blank space, optional minus sign, and zero or more digits (including decimal point) is sorted by arithmetic value. Leading blank characters are ignored.
, --random-sort, --sort=random
Sort lines in random order. This is a random permutation of the inputs with the exception that equal keys sort together. It is implemented by hashing the input keys and sorting the hash values. The hash function is randomized with data from arc4random_buf(3), or by file content if one is specified via --random-source. If multiple sort fields are specified, the same random hash function is used for all of them.
, --reverse
Sort in reverse order.
, --version-sort
Sort version numbers. The input lines are treated as file names in form PREFIX VERSION SUFFIX, where SUFFIX matches the regular expression "(.([A-Za-z~][A-Za-z0-9~]*)?)*". The files are compared by their prefixes and versions (leading zeros are ignored in version numbers, see example below). If an input string does not match the pattern, then it is compared using the byte compare function.

For example:

$ ls sort* | sort -V

The treatment of field separators can be altered using these options:

, --ignore-leading-blanks
Ignore leading blank space when determining the start and end of a restricted sort key (see -k). If -b is specified before the first -k option, it applies globally to all key specifications. Otherwise, -b can be attached independently to each field argument of the key specifications. Note that -b should not appear after -k, and that it has no effect unless key fields are specified.
field1[,field2], --key=field1[,field2]
Define a restricted sort key that has the starting position field1, and optional ending position field2 of a key field. The -k option may be specified multiple times, in which case subsequent keys are compared after earlier keys compare equal. The -k option replaces the obsolete options +pos1 and -pos2, but the old notation is also supported.
char, --field-separator=char
Use char as the field separator character. The initial char is not considered to be part of a field when determining key offsets. Each occurrence of char is significant (for example, “charchar” delimits an empty field). If -t is not specified, the default field separator is a sequence of blank-space characters, and consecutive blank spaces do delimit an empty field; further, the initial blank space considered part of a field when determining key offsets. To use NUL as field separator, use -t '\0'.
, --zero-terminated
Use NUL as the record separator. By default, records in the files are expected to be separated by the newline characters. With this option, NUL (‘\0’) is used as the record separator character.

Other options:

Specify maximum number of files that can be opened by sort at once. This option affects behavior when having many input files or using temporary files. The minimum value is 2. The default value is 16.
Use program to compress temporary files. When invoked with no arguments, program must compress standard input to standard output. When called with the -d option, it must decompress standard input to standard output. If program fails, sort will exit with an error. The compress(1) and gzip(1) utilities meet these requirements.
Print some extra information about the sorting process to the standard output.
Take the input file list from the file filename. The file names must be separated by NUL (like the output produced by the command “find ... -print0”).
Try to use heap sort, if the sort specifications allow. This sort algorithm cannot be used with -u and -s.
Print the help text and exit.
, --mergesort
Use mergesort. This is a universal algorithm that can always be used, but it is not always the fastest.
Try to use file memory mapping system call. It may increase speed in some cases.
Try to use quick sort, if the sort specifications allow. This sort algorithm cannot be used with -u and -s.
Try to use radix sort, if the sort specifications allow. The radix sort can only be used for trivial locales (C and POSIX), and it cannot be used for numeric or month sort. Radix sort is very fast and stable.
For random sort, the contents of filename are used as the source of the ‘seed’ data for the hash function. Two invocations of random sort with the same seed data produce the same result if the input is also identical. By default, the arc4random_buf(3) function is used instead.
Print the version and exit.

A field is defined as a maximal sequence of characters other than the field separator and record separator (newline by default). Initial blank spaces are included in the field unless -b has been specified; the first blank space of a sequence of blank spaces acts as the field separator and is included in the field (unless -t is specified). For example, by default all blank spaces at the beginning of a line are considered to be part of the first field.

Fields are specified by the -k field1[,field2] option. If field2 is missing, the end of the key defaults to the end of the line.

The arguments field1 and field2 have the form m.n and can be followed by one or more of the modifiers b, d, f, i, n, g, M and r, which correspond to the options discussed above. When b is specified, it applies only to field1 or field2 where it is specified while the rest of the modifiers apply to the whole key field regardless if they are specified only with field1 or field2 or both. A field1 position specified by m.n is interpreted as the nth character from the beginning of the mth field. A missing .n in field1 means ‘.1’, indicating the first character of the mth field; if the -b option is in effect, n is counted from the first non-blank character in the mth field; m.1b refers to the first non-blank character in the mth field. 1.n refers to the nth character from the beginning of the line; if n is greater than the length of the line, the field is taken to be empty.

nth positions are always counted from the field beginning, even if the field is shorter than the number of specified positions. Thus, the key can really start from a position in a subsequent field.

A field2 position specified by m.n is interpreted as the nth character (including separators) from the beginning of the mth field. A missing .n indicates the last character of the mth field; m = 0 designates the end of a line. Thus the option -k v.x,w.y is synonymous with the obsolete option +v-1.x-1 -w-1.y; when is omitted, -k v.x,w is synonymous with +v-1.x-1 -w.0. The obsolete +pos1 -pos2 option is still supported, except for -w.0b, which has no -k equivalent.

Path to the directory in which temporary files will be stored. Note that TMPDIR may be overridden by the -T option.

Temporary files.

The sort utility exits with one of the following values:

Successfully sorted the input files or if used with -C or -c, the input file already met the sorting criteria.
On disorder (or non-uniqueness) with the -C or -c options.
An error occurred.

comm(1), join(1), uniq(1)

The sort utility is compliant with the IEEE Std 1003.1-2008 (“POSIX.1”) specification, except that it ignores the user's locale(1) and always assumes LC_ALL=C.

The flags [-gHhiMRSsTVz] are extensions to that specification.

All long options are extensions to the specification. Some are provided for compatibility with GNU sort, others are specific to this implementation.

Some implementations of sort honor the -b option even when no key fields are specified. This implementation follows historic practice and IEEE Std 1003.1-2008 (“POSIX.1”) in only honoring -b when it precedes a key field.

The historic practice of allowing the -o option to appear after the file is supported for compatibility with older versions of sort.

The historic key notations +pos1 and -pos2 are supported for compatibility with older versions of sort but their use is highly discouraged.

A sort command appeared in Version 1 AT&T UNIX.

Gabor Kovesdan <>
Oleg Moskalenko <>

This implementation of sort has no limits on input line length (other than imposed by available memory) or any restrictions on bytes allowed within lines.

The performance depends highly on efficient choice of sort keys and key complexity. The fastest sort is on whole lines, with option -s. For the key specification, the simpler to process the lines the faster the search will be.

When sorting by arithmetic value, using -n results in much better performance than -g so its use is encouraged whenever possible.

March 31, 2022 OpenBSD-7.3