xref: /openbsd/usr.bin/sort/sort.1 (revision 41ce3b17)
1.\"	$OpenBSD: sort.1,v 1.65 2022/03/31 17:27:27 naddy Exp $
2.\"
3.\" Copyright (c) 1991, 1993
4.\"	The Regents of the University of California.  All rights reserved.
5.\"
6.\" This code is derived from software contributed to Berkeley by
7.\" the Institute of Electrical and Electronics Engineers, Inc.
8.\"
9.\" Redistribution and use in source and binary forms, with or without
10.\" modification, are permitted provided that the following conditions
11.\" are met:
12.\" 1. Redistributions of source code must retain the above copyright
13.\"    notice, this list of conditions and the following disclaimer.
14.\" 2. Redistributions in binary form must reproduce the above copyright
15.\"    notice, this list of conditions and the following disclaimer in the
16.\"    documentation and/or other materials provided with the distribution.
17.\" 3. Neither the name of the University nor the names of its contributors
18.\"    may be used to endorse or promote products derived from this software
19.\"    without specific prior written permission.
20.\"
21.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31.\" SUCH DAMAGE.
32.\"
33.\"     @(#)sort.1	8.1 (Berkeley) 6/6/93
34.\"
35.Dd $Mdocdate: March 31 2022 $
36.Dt SORT 1
37.Os
38.Sh NAME
39.Nm sort
40.Nd sort, merge, or sequence check text and binary files
41.Sh SYNOPSIS
42.Nm sort
43.Op Fl bCcdfgHhiMmnRrsuVz
44.Op Fl k Ar field1 Ns Op , Ns Ar field2
45.Op Fl o Ar output
46.Op Fl S Ar size
47.Op Fl T Ar dir
48.Op Fl t Ar char
49.Op Ar
50.Sh DESCRIPTION
51The
52.Nm
53utility sorts text and binary files by lines.
54A line is a record separated from the subsequent record by a
55newline (default) or NUL
56.Ql \e0
57character
58.Po
59.Fl z
60option
61.Pc .
62A record can contain any printable or unprintable characters.
63Comparisons are based on one or more sort keys extracted from
64each line of input, and are performed lexicographically,
65according to the specified command-line options
66that can tune the actual sorting behavior.
67By default, if keys are not given,
68.Nm
69uses entire lines for comparison.
70.Pp
71If no
72.Ar file
73is specified, or if
74.Ar file
75is
76.Sq - ,
77the standard input is used.
78.Pp
79The options are as follows:
80.Bl -tag -width Ds
81.It Fl C , Fl Fl check Ns = Ns Cm silent Ns | Ns Cm quiet
82Check that the single input file is sorted.
83If it is, exit 0; if it's not, exit 1.
84In either case, produce no output.
85.It Fl c , Fl Fl check
86Like
87.Fl C ,
88but additionally write a message to
89.Em stderr
90if the input file is not sorted.
91.It Fl m , Fl Fl merge
92Merge only; the input files are assumed to be pre-sorted.
93If they are not sorted, the output order is undefined.
94.It Fl o Ar output , Fl Fl output Ns = Ns Ar output
95Write the output to the
96.Ar output
97file instead of the standard output.
98This file can be the same as one of the input files.
99.It Fl S Ar size , Fl Fl buffer-size Ns = Ns Ar size
100Use a memory buffer no larger than
101.Ar size .
102The modifiers %, b, K, M, G, T, P, E, Z, and Y can be used.
103If no memory limit is specified,
104.Nm
105may use up to about 90% of available memory.
106If the input is too big to fit into the memory buffer,
107temporary files are used.
108.It Fl s
109Stable sort; maintains the original record order of records that have
110an equal key.
111This is a non-standard feature, but it is widely accepted and used.
112.It Fl T Ar dir , Fl Fl temporary-directory Ns = Ns Ar dir
113Store temporary files in the directory
114.Ar dir .
115The default path is the value of the environment variable
116.Ev TMPDIR
117or
118.Pa /tmp
119if
120.Ev TMPDIR
121is not defined.
122.It Fl u , Fl Fl unique
123Unique: suppress all but one in each set of lines having equal keys.
124This option implies a stable sort (see below).
125If used with
126.Fl C
127or
128.Fl c ,
129.Nm
130also checks that there are no lines with duplicate keys.
131.El
132.Pp
133The following options override the default ordering rules.
134If ordering options appear before the first
135.Fl k
136option, they apply globally to all sort keys.
137When attached to a specific key (see
138.Fl k ) ,
139the ordering options override all global ordering options for that key.
140Note that the ordering options intended to apply globally should not
141appear after
142.Fl k
143or results may be unexpected.
144.Bl -tag -width indent
145.It Fl d , Fl Fl dictionary-order
146Consider only blank spaces and alphanumeric characters in comparisons.
147.It Fl f , Fl Fl ignore-case
148Consider all lowercase characters that have uppercase
149equivalents to be the same for purposes of comparison.
150.It Fl g , Fl Fl general-numeric-sort , Fl Fl sort Ns = Ns Cm general-numeric
151Sort by general numerical value.
152As opposed to
153.Fl n ,
154this option handles general floating points.
155It has a more
156permissive format than that allowed by
157.Fl n
158but it has a significant performance drawback.
159.It Fl h , Fl Fl human-numeric-sort , Fl Fl sort Ns = Ns Cm human-numeric
160Sort by numerical value, but take into account the SI suffix,
161if present.
162Sorts first by numeric sign (negative, zero, or
163positive); then by SI suffix (either empty, or `k' or `K', or one
164of `MGTPEZY', in that order); and finally by numeric value.
165The SI suffix must immediately follow the number.
166For example, '12345K' sorts before '1M', because M is "larger" than K.
167This sort option is useful for sorting the output of a single invocation
168of 'df' command with
169.Fl h
170or
171.Fl H
172options (human-readable).
173.It Fl i , Fl Fl ignore-nonprinting
174Ignore all non-printable characters.
175.It Fl M , Fl Fl month-sort , Fl Fl sort Ns = Ns Cm month
176Sort by month abbreviations.
177Unknown strings are considered smaller than valid month names.
178.It Fl n , Fl Fl numeric-sort , Fl Fl sort Ns = Ns Cm numeric
179An initial numeric string, consisting of optional blank space, optional
180minus sign, and zero or more digits (including decimal point)
181is sorted by arithmetic value.
182Leading blank characters are ignored.
183.It Fl R , Fl Fl random-sort , Fl Fl sort Ns = Ns Cm random
184Sort lines in random order.
185This is a random permutation of the inputs with the exception that
186equal keys sort together.
187It is implemented by hashing the input keys and sorting the hash values.
188The hash function is randomized with data from
189.Xr arc4random_buf 3 ,
190or by file content if one is specified via
191.Fl Fl random-source .
192If multiple sort fields are specified,
193the same random hash function is used for all of them.
194.It Fl r , Fl Fl reverse
195Sort in reverse order.
196.It Fl V , Fl Fl version-sort
197Sort version numbers.
198The input lines are treated as file names in form
199PREFIX VERSION SUFFIX, where SUFFIX matches the regular expression
200"(\.([A-Za-z~][A-Za-z0-9~]*)?)*".
201The files are compared by their prefixes and versions (leading
202zeros are ignored in version numbers, see example below).
203If an input string does not match the pattern, then it is compared
204using the byte compare function.
205.Pp
206For example:
207.Bd -literal -offset indent
208$ ls sort* | sort -V
209sort-1.022.tgz
210sort-1.23.tgz
211sort-1.23.1.tgz
212sort-1.024.tgz
213sort-1.024.003.
214sort-1.024.003.tgz
215sort-1.024.07.tgz
216sort-1.024.009.tgz
217.Ed
218.El
219.Pp
220The treatment of field separators can be altered using these options:
221.Bl -tag -width indent
222.It Fl b , Fl Fl ignore-leading-blanks
223Ignore leading blank space when determining the start
224and end of a restricted sort key (see
225.Fl k ) .
226If
227.Fl b
228is specified before the first
229.Fl k
230option, it applies globally to all key specifications.
231Otherwise,
232.Fl b
233can be attached independently to each
234.Ar field
235argument of the key specifications.
236Note that
237.Fl b
238should not appear after
239.Fl k ,
240and that it has no effect unless key fields are specified.
241.It Xo
242.Fl k Ar field1 Ns Op , Ns Ar field2 ,
243.Fl Fl key Ns = Ns Ar field1 Ns Op , Ns Ar field2
244.Xc
245Define a restricted sort key that has the starting position
246.Ar field1 ,
247and optional ending position
248.Ar field2
249of a key field.
250The
251.Fl k
252option may be specified multiple times,
253in which case subsequent keys are compared after earlier keys compare equal.
254The
255.Fl k
256option replaces the obsolete options
257.Cm \(pl Ns Ar pos1
258and
259.Fl Ns Ar pos2 ,
260but the old notation is also supported.
261.It Fl t Ar char , Fl Fl field-separator Ns = Ns Ar char
262Use
263.Ar char
264as the field separator character.
265The initial
266.Ar char
267is not considered to be part of a field when determining key offsets.
268Each occurrence of
269.Ar char
270is significant (for example,
271.Dq Ar charchar
272delimits an empty field).
273If
274.Fl t
275is not specified, the default field separator is a sequence of
276blank-space characters, and consecutive blank spaces do
277.Em not
278delimit an empty field; further, the initial blank space
279.Em is
280considered part of a field when determining key offsets.
281To use NUL as field separator, use
282.Fl t
283\(aq\e0\(aq.
284.It Fl z , Fl Fl zero-terminated
285Use NUL as the record separator.
286By default, records in the files are expected to be separated by
287the newline characters.
288With this option, NUL
289.Pq Ql \e0
290is used as the record separator character.
291.El
292.Pp
293Other options:
294.Bl -tag -width indent
295.It Fl Fl batch-size Ns = Ns Ar num
296Specify maximum number of files that can be opened by
297.Nm
298at once.
299This option affects behavior when having many input files or using
300temporary files.
301The minimum value is 2.
302The default value is 16.
303.It Fl Fl compress-program Ns = Ns Ar program
304Use
305.Ar program
306to compress temporary files.
307When invoked with no arguments,
308.Ar program
309must compress standard input to standard output.
310When called with the
311.Fl d
312option, it must decompress standard input to standard output.
313If
314.Ar program
315fails,
316.Nm
317will exit with an error.
318The
319.Xr compress 1
320and
321.Xr gzip 1
322utilities meet these requirements.
323.It Fl Fl debug
324Print some extra information about the sorting process to the
325standard output.
326.It Fl Fl files0-from Ns = Ns Ar filename
327Take the input file list from the file
328.Ar filename .
329The file names must be separated by NUL
330(like the output produced by the command
331.Dq find ... -print0 ) .
332.It Fl Fl heapsort
333Try to use heap sort, if the sort specifications allow.
334This sort algorithm cannot be used with
335.Fl u
336and
337.Fl s .
338.It Fl Fl help
339Print the help text and exit.
340.It Fl H , Fl Fl mergesort
341Use mergesort.
342This is a universal algorithm that can always be used,
343but it is not always the fastest.
344.It Fl Fl mmap
345Try to use file memory mapping system call.
346It may increase speed in some cases.
347.It Fl Fl qsort
348Try to use quick sort, if the sort specifications allow.
349This sort algorithm cannot be used with
350.Fl u
351and
352.Fl s .
353.It Fl Fl radixsort
354Try to use radix sort, if the sort specifications allow.
355The radix sort can only be used for trivial locales (C and POSIX),
356and it cannot be used for numeric or month sort.
357Radix sort is very fast and stable.
358.It Fl Fl random-source Ns = Ns Ar filename
359For random sort, the contents of
360.Ar filename
361are used as the source of the
362.Sq seed
363data for the hash function.
364Two invocations of random sort with the same seed data
365produce the same result if the input is also identical.
366By default, the
367.Xr arc4random_buf 3
368function is used instead.
369.It Fl Fl version
370Print the version and exit.
371.El
372.Pp
373A field is defined as a maximal sequence of characters other than the
374field separator and record separator
375.Pq newline by default .
376Initial blank spaces are included in the field unless
377.Fl b
378has been specified;
379the first blank space of a sequence of blank spaces acts as the field
380separator and is included in the field (unless
381.Fl t
382is specified).
383For example, by default all blank spaces at the beginning of a line are
384considered to be part of the first field.
385.Pp
386Fields are specified by the
387.Fl k Ar field1 Ns Op , Ns Ar field2
388option.
389If
390.Ar field2
391is missing, the end of the key defaults to the end of the line.
392.Pp
393The arguments
394.Ar field1
395and
396.Ar field2
397have the form
398.Em m.n
399.Em (m,n > 0)
400and can be followed by one or more of the modifiers
401.Cm b , d , f , i ,
402.Cm n , g , M
403and
404.Cm r ,
405which correspond to the options discussed above.
406When
407.Cm b
408is specified, it applies only to
409.Ar field1
410or
411.Ar field2
412where it is specified while the rest of the modifiers
413apply to the whole key field regardless if they are
414specified only with
415.Ar field1
416or
417.Ar field2
418or both.
419A
420.Ar field1
421position specified by
422.Em m.n
423is interpreted as the
424.Em n Ns th
425character from the beginning of the
426.Em m Ns th
427field.
428A missing
429.Em \&.n
430in
431.Ar field1
432means
433.Ql \&.1 ,
434indicating the first character of the
435.Em m Ns th
436field; if the
437.Fl b
438option is in effect,
439.Em n
440is counted from the first non-blank character in the
441.Em m Ns th
442field;
443.Em m Ns \&.1b
444refers to the first non-blank character in the
445.Em m Ns th
446field.
447.No 1\&. Ns Em n
448refers to the
449.Em n Ns th
450character from the beginning of the line;
451if
452.Em n
453is greater than the length of the line, the field is taken to be empty.
454.Pp
455.Em n Ns th
456positions are always counted from the field beginning, even if the field
457is shorter than the number of specified positions.
458Thus, the key can really start from a position in a subsequent field.
459.Pp
460A
461.Ar field2
462position specified by
463.Em m.n
464is interpreted as the
465.Em n Ns th
466character (including separators) from the beginning of the
467.Em m Ns th
468field.
469A missing
470.Em \&.n
471indicates the last character of the
472.Em m Ns th
473field;
474.Em m
475= \&0
476designates the end of a line.
477Thus the option
478.Fl k Ar v.x,w.y
479is synonymous with the obsolete option
480.Cm \(pl Ns Ar v-\&1.x-\&1
481.Fl Ns Ar w-\&1.y ;
482when
483.Em y
484is omitted,
485.Fl k Ar v.x,w
486is synonymous with
487.Cm \(pl Ns Ar v-\&1.x-\&1
488.Fl Ns Ar w\&.0 .
489The obsolete
490.Cm \(pl Ns Ar pos1
491.Fl Ns Ar pos2
492option is still supported, except for
493.Fl Ns Ar w\&.0b ,
494which has no
495.Fl k
496equivalent.
497.Sh ENVIRONMENT
498.Bl -tag -width Ds
499.It Ev TMPDIR
500Path to the directory in which temporary files will be stored.
501Note that
502.Ev TMPDIR
503may be overridden by the
504.Fl T
505option.
506.El
507.Sh FILES
508.Bl -tag -width Pa -compact
509.It Pa /tmp/.bsdsort.PID.*
510Temporary files.
511.El
512.Sh EXIT STATUS
513The
514.Nm
515utility exits with one of the following values:
516.Pp
517.Bl -tag -width Ds -offset indent -compact
518.It 0
519Successfully sorted the input files or if used with
520.Fl C
521or
522.Fl c ,
523the input file already met the sorting criteria.
524.It 1
525On disorder (or non-uniqueness) with the
526.Fl C
527or
528.Fl c
529options.
530.It 2
531An error occurred.
532.El
533.Sh SEE ALSO
534.Xr comm 1 ,
535.Xr join 1 ,
536.Xr uniq 1
537.Sh STANDARDS
538The
539.Nm
540utility is compliant with the
541.St -p1003.1-2008
542specification, except that it ignores the user's
543.Xr locale 1
544and always assumes
545.Ev LC_ALL Ns =C.
546.Pp
547The flags
548.Op Fl gHhiMRSsTVz
549are extensions to that specification.
550.Pp
551All long options are extensions to the specification.
552Some are provided for compatibility with GNU
553.Nm ,
554others are specific to this implementation.
555.Pp
556Some implementations of
557.Nm
558honor the
559.Fl b
560option even when no key fields are specified.
561This implementation follows historic practice and
562.St -p1003.1-2008
563in only honoring
564.Fl b
565when it precedes a key field.
566.Pp
567The historic practice of allowing the
568.Fl o
569option to appear after the
570.Ar file
571is supported for compatibility with older versions of
572.Nm .
573.Pp
574The historic key notations
575.Cm \(pl Ns Ar pos1
576and
577.Fl Ns Ar pos2
578are supported for compatibility with older versions of
579.Nm
580but their use is highly discouraged.
581.Sh HISTORY
582A
583.Nm
584command appeared in
585.At v1 .
586.Sh AUTHORS
587.An Gabor Kovesdan Aq Mt gabor@FreeBSD.org
588.An Oleg Moskalenko Aq Mt mom040267@gmail.com
589.Sh CAVEATS
590This implementation of
591.Nm
592has no limits on input line length (other than imposed by available
593memory) or any restrictions on bytes allowed within lines.
594.Pp
595The performance depends highly on
596efficient choice of sort keys and key complexity.
597The fastest sort is on whole lines, with option
598.Fl s .
599For the key specification, the simpler to process the
600lines the faster the search will be.
601.Pp
602When sorting by arithmetic value, using
603.Fl n
604results in much better performance than
605.Fl g
606so its use is encouraged whenever possible.
607