xref: /openbsd-src/regress/usr.bin/sort/stests (revision 4006d63ad31f284fd33affeb94493e35cabac953)

#!/bin/sh -
#	$OpenBSD: stests,v 1.2 2001/02/04 21:29:31 ericj Exp $
#	from: @(#)stests	8.1 (Berkeley) 6/6/93

#Latest version.  My sort passes all tests because I wrote it.
#We differ only on 25E and 25H.
#(I found at least one bug in constructing test 25, and was driven
#to rewrite field parsing to clarify it.)
#
#In 25E, -k2.3,2.1b, the fields are not necessarily out of order.
#Even if they were, it would be legal (11752-3), although certainly
#justification for warning.
#
#On 25H, your answer is as defensible as mine.  (Our suggestion
#*1 backs mine.)


# Tests for the Unix sort utility
# Test Posix features except for locale.
# Test some nonstandard features if present.

# Other tests should be made for files too big to fit in memory.


# Initialize switches for nonstandard features.
# Use parenthesized settings for supported features.

o=:	# officially obsolescent features: +1 -2, misplaced -o (o=)
g=:	# -g numeric sort including e-format numbers (g=)
M=:	# -M sort by month names (M=)
s=:	# -s stable, do not compare raw bytes on equal keys (s=)
y=	# -y user-specified memory size (y=-y10000)

# Detect what features are supported, assuming bad options cause
# errors.  Set switches accordingly.

echo obsolescent and nonstandard features recognized, if any:
if sort +0 </dev/null 2>/dev/null; then o=
				echo '	+1 -2'; fi
if sort /dev/null -o xx 2>/dev/null; then o=
				echo '	displaced -o'; fi
if sort -g </dev/null 2>/dev/null; then g=
				echo '	-g g-format numbers'; fi
if sort -M </dev/null 2>/dev/null; then M=
				echo '	-M months'; fi
if sort -s </dev/null 2>/dev/null; then s=
				echo '	-s stable'; fi
if sort -y10000 </dev/null 2>/dev/null; then y=-y10000
				echo '	-y space'; fi
if sort -z10000 </dev/null 2>/dev/null; then
				echo '	-z size (not exercised)'; fi
if sort -T. </dev/null 2>/dev/null; then
				echo '	-T tempdir (not exercised)'; fi


export TEST	# major sequence number of test

trap "rm -f in in1 out xx -k xsort linecount fields; exit" 0 1 2 13 15

# xsort testno options
# Sort file "in" with specified options.
# Compare with file "out" if that is supplied,
# otherwise make plausibility checks on output

#	"sum" must be dumb; insensitive to the
#	order of lines within a file.
#	System V sum is suitable; sum -5 is the v10 equivalent.

PATH=.:$PATH
export PATH
cat <<'!' >xsort; chmod +x xsort

	X=$1; shift

	if sort "$@" in >xx  &&  sort -c "$@" xx
	then
		if test -f out
		then
			cmp xx out >/dev/null && exit 0
			echo $TEST$X comparison failed
		else
			test "`cksum -o2 <in`" = "`cksum -o2 <xx`" && exit 0
			echo $TEST$X checksum failed
		fi
	else
		echo $TEST$X failed
	fi
	exit 1
!

# linecount testno file count
# declares the given "testno" to be in error if number of
# lines in "file" differs from "count"

cat <<'!' >linecount; chmod +x linecount
awk 'END{ if(NR!='$3') print "'$TEST$1' failed" }' $2
!

rm -f out

#---------------------------------------------------------------
TEST=01; echo $TEST	# -c status, checksum
			# obsolescent features go together
cat <<! >in
b
a
!
rm -f out -o

sort -c in 2>/dev/null && echo ${TEST}A failed

xsort B || '"cksum"' is probably unsuitable - see comments

$o sort +0 in -o in || echo ${TEST}c failed

#---------------------------------------------------------------
TEST=02; echo $TEST	# output from -c
cat <<! >in
x
y
!

sort -cr in >out 2>xx && echo ${TEST}A failed
test -s out && echo ${TEST}B failed
test -s xx && echo option -c is noisy "(probably legal)"
test -s xx || echo option -c is quiet "(legal, not classical)"

#---------------------------------------------------------------
TEST=03; echo $TEST	# -n
cat <<! >in
-99.0
-99.1
-.0002
-10
2
0010.000000000000000000000000000000000001
10
3x
x
!
cat <<! >out
-99.1
-99.0
-10
-.0002
x
2
3x
10
0010.000000000000000000000000000000000001
!

xsort "" -n

#---------------------------------------------------------------
TEST=04; echo $TEST	# -b without fields, piping, -c status return
cat <<! >in
  b
 a
!
cp in out

xsort A -b

cat in | sort | cat >xx
cmp xx out >/dev/null || echo ${TEST}B failed

sort in | sort -cr 2>/dev/null && echo ${TEST}C failed

#---------------------------------------------------------------
TEST=05; echo $TEST	# fields, reverse fields, -c status return
cat <<! >in
b b p
a b q
x a
!
cat <<! >out
x a
a b q
b b p
!

$o xsort A +1 -2

$o xsort B +1 -2 +2r

xsort C -k 2,2

xsort D -k 2,2 -k 3r

xsort E -k 2,2.0

xsort F -k 2,2 -k 1,1 -k 3

sort -c -k 2 in 2>/dev/null && echo ${TEST}G failed

#---------------------------------------------------------------
TEST=06; echo $TEST	# -t
cat <<! >in
a:
a!
!
cp in out

$o xsort A -t : -r +0

$o xsort B -t : +0 -1

xsort C -t : -r -k 1

xsort D -t : -k 1,1

#---------------------------------------------------------------
TEST=07; echo $TEST	# -t, character positions in fields
	# -t: as 1 arg is not strictly conforming, but classical
cat <<! >in
: ab
:bac
!
cat <<! >out
:bac
: ab
!

$o xsort A -b -t: +1.1

$o xsort B -t: +1.1r

xsort C -b -t: -k 2.2

xsort D -t: -k 2.2r

#---------------------------------------------------------------
TEST=08; echo $TEST	# space and tab as -t characters
cat <<! >in
 b c
 b	c
	b c
!
cp in out

xsort A -t ' ' -k2,2

xsort B -t ' ' -k2.1,2.0

cat <<! >out
 b c
	b c
 b	c
!

xsort C -t '	' -k2,2

xsort D -t '	' -k2.1,2.0

cat <<! >out
 b	c
	b c
 b c
!

xsort E -k2

cat <<! >out
	b c
 b	c
 b c
!

xsort F -k2b

#---------------------------------------------------------------
TEST=09; echo $TEST	# alphabetic as -t character
cat <<! >in
zXa
yXa
zXb
!
cp in out

xsort "" -tX -k2 -k1r,1

#---------------------------------------------------------------
TEST=10; echo $TEST	# -m
cat <<! >in
a
ab
ab
bc
ca
!
cat <<! >in1
Z
a
aa
ac
c
!
cat <<! >out
Z
a
a
aa
ab
ab
ac
bc
c
ca
!

sort -m in in1 >xx
cmp xx out >/dev/null || echo $TEST failed

#---------------------------------------------------------------
TEST=11; echo $TEST	# multiple files, -o overwites input, -m, -mu
cat <<! >in
a
b
c
d
!

sort -o xx     in in in in in in in in in in in in in in in in in
linecount A xx 68
sort -o in -mu in in in in in in in in in in in in in in in in in
linecount B in 4
sort -o in -m  in in in in in in in in in in in in in in in in in

cmp in xx >/dev/null || echo ${TEST}C failed

#---------------------------------------------------------------
TEST=12; echo $TEST	# does -mu pick the first among equals?
cat <<! >in
3B
3b
3B2
~3B2
4.1
41
5
5.
!
cat <<! >out
3B
3B2
4.1
5
!

xsort A -mudf || echo "(other behavior is legal, not classical)"

xsort B -mudf -k1 || echo "(other behavior is legal, not classical)"

#---------------------------------------------------------------
TEST=13; echo $TEST	# long records (>8000 bytes, keys >16000), -r
awk '
BEGIN {	x="x"
	for(i=1; i<=12; i++) x = x x
	for(i=15; i<=25; i++) print x i
}' >in
awk '
BEGIN {	x="x"
	for(i=1; i<=12; i++) x = x x
	for(i=25; i>=15; i--) print x i
}' >out

xsort A -r

xsort B -k 1,1r -k 1

#---------------------------------------------------------------
TEST=14; echo $TEST "(3 long parts)"
awk 'BEGIN { for(i=0; i<100000; i++) print rand() }' | grep -v e >in
rm -f out

xsort A; echo $TEST "(part A done)"

xsort B -n; echo $TEST "(part B done)"

# next test is unclean: xx is a hidden side-effect of xsort

awk '
	$0 < x { print "test '${TEST}C' failed"; exit }
	$0 "" != x { print >"out"; x = $0 }
' xx

xsort C -n -u

#---------------------------------------------------------------
TEST=15; echo $TEST "(long)"	# force intermediate files if possible
awk 'BEGIN { for(i=0; i<20000; i++) print rand() }' >in
rm -f out

xsort A -r $y

sort -r in | awk '$0 "x" != x { print ; x = $0 "x" }' >out

xsort B -u -r $y

#---------------------------------------------------------------
TEST=16; echo $TEST	# -nr, -nm, file name -
awk 'BEGIN { for(i=-100; i<=100; i+=2) printf "%.10d\n", i }' >in

awk 'BEGIN { for(i=-99; i<=100; i+=2) print i }' | sort -nr in - >xx
awk '$0+0 != 101-NR { print "'${TEST}A' failed"; exit }' xx

awk 'BEGIN { for(i=-99; i<=100; i+=2) print i }' | sort -mn in - >xx
awk '$0+0 != -101+NR { print "'${TEST}B' failed"; exit }' xx

#---------------------------------------------------------------
TEST=17; echo $TEST	# -d, fields without end, modifier override
cat <<! >in
a-B
a+b
a b
A+b
a	b
!
cat <<! >out
a	b
a b
A+b
a-B
a+b
!

$o xsort A -df +0 +0d

xsort B -df -k 1 -k 1d

#---------------------------------------------------------------
TEST=18; echo $TEST	# -u on key only
cat <<! >in
12	y
13	z
12	x
!
cat <<! >out
12	x
12	y
13	z
!

$o xsort A +0 -1

xsort B -k 1,1

sort -u -k 1,1 in >xx
linecount C xx 2

#---------------------------------------------------------------
TEST=19; echo $TEST	# -i, -d, -f
cat <<! >xx.c
run(i,j){ for( ; i<=j; i++) printf("%.3o %c\n",i,i); }
main(){	run(0, 011);		/* 012=='\n' */
	run(013, 0377); }
!
cc xx.c
a.out >in
cat <<! >xx.c
run(i,j){ for( ; i<=j; i++) printf("%.3o %c\n",i,i); }
main(){ run(0, 011);
	run(013, ' '-1);
	run(0177, 0377);
	run(' ', 0176); }
!
cc xx.c
a.out >out

xsort A -i -k 2

cat <<! >xx.c
run(i,j){ for( ; i<=j; i++) printf("%.3o %c\n",i,i); }
main(){	run(0, 010);		/* 011=='\t', 012=='\n' */
	run(013, ' '-1);
	run(' '+1, '0'-1);
	run('9'+1, 'A'-1);
	run('Z'+1, 'a'-1);
	run('z'+1, 0377);
	run('\t', '\t');
	run(' ', ' ');
	run('0', '9');
	run('A', 'Z');
	run('a', 'z'); }
!
cc xx.c
a.out >out

xsort B -d -k 2

cat <<! >xx.c
run(i,j){ for( ; i<=j; i++) printf("%.3o %c\n",i,i); }
main(){	int i;
	run(0, 011);
	run(013, 'A'-1);
	for(i='A'; i<='Z'; i++)
		printf("%.3o %c\n%.3o %c\n",i,i,i+040,i+040);
	run('Z'+1, 'a'-1);
	run('z'+1, 0377); }
!
cc xx.c
a.out >out
rm xx.c

xsort C -f -k 2

#---------------------------------------------------------------
TEST=20; echo $TEST	# -d, -f, -b applies only to fields
cat <<! >in
 b
'C
a
!
cp in out

xsort A -d

xsort B -f

cat <<! >out
 b
a
'C
!

xsort C -dfb

#---------------------------------------------------------------
TEST=21; echo $TEST	# behavior of null bytes
cat <<'!' >xx.c
main() { printf("%cb\n%ca\n",0,0); }
!
cc xx.c
a.out >in
sort in >xx
cmp in xx >/dev/null && echo ${TEST}A failed
test "`wc -c <in`" = "`wc -c <xx`" || echo ${TEST}B failed
rm xx.c a.out

#---------------------------------------------------------------
TEST=22; echo $TEST	# field limits
cat <<! >in
a	2
a	1
b	2
b	1
!
cat <<! >out
b	1
b	2
a	1
a	2
!

xsort "" -r -k1,1 -k2n

#---------------------------------------------------------------
TEST=23; echo $TEST	# empty file

sort -o xx </dev/null
cmp xx /dev/null 2>/dev/null || echo ${TEST}A failed

sort -c </dev/null || echo ${TEST}B failed

sort -cu </dev/null || echo ${TEST}C failed

#---------------------------------------------------------------
TEST=24; echo $TEST	# many fields
cat <<! >in
0:2:3:4:5:6:7:8:9
1:1:3:4:5:6:7:8:9
1:2:2:4:5:6:7:8:9
1:2:3:3:5:6:7:8:9
1:2:3:4:4:6:7:8:9
1:2:3:4:5:5:7:8:9
1:2:3:4:5:6:6:8:9
1:2:3:4:5:6:7:7:9
1:2:3:4:5:6:7:8:8
!
cat <<! >out
1:2:3:4:5:6:7:8:8
1:2:3:4:5:6:7:7:9
1:2:3:4:5:6:6:8:9
1:2:3:4:5:5:7:8:9
1:2:3:4:4:6:7:8:9
1:2:3:3:5:6:7:8:9
1:2:2:4:5:6:7:8:9
1:1:3:4:5:6:7:8:9
0:2:3:4:5:6:7:8:9
!

xsort "" -t: -k9 -k8 -k7 -k6 -k5 -k4 -k3 -k2 -k1

#---------------------------------------------------------------
TEST=25; echo $TEST	# variously specified alpha fields
			# numbers give the correct orderings
cat <<! >in
01:04:19:01:16:01:21:01 a
02:03:13:15:13:19:15:02  a
03:02:07:09:07:13:09:03   a
04:01:01:03:01:07:03:04    a
05:08:20:16:17:02:20:05 aa
06:07:14:18:14:20:14:06  aa
07:06:08:10:08:14:08:07   aa
08:05:02:04:02:08:02:08    aa
09:16:22:02:22:04:24:13 b
10:15:16:20:19:22:18:14  b
11:14:10:12:10:16:12:15   b
12:13:04:06:04:10:06:16    b
13:24:24:22:24:06:22:21 bb
14:23:18:24:21:24:16:22  bb
15:22:12:14:12:18:10:23   bb
16:21:06:08:06:12:04:24    bb
17:12:21:21:18:03:19:09 ab
18:11:15:19:15:21:13:10  ab
19:10:09:11:09:15:07:11   ab
20:09:03:05:03:09:01:12    ab
21:20:23:17:23:05:23:17 ba
22:19:17:23:20:23:17:18  ba
23:18:11:13:11:17:11:19   ba
24:17:05:07:05:11:05:20    ba
!
sort -k2b -k2 in >xx &&
	sort -c -t: -k2n xx 2>/dev/null || echo ${TEST}A failed
sort -k2,2.1b -k2 in >xx &&
	sort -c -t: -k3n xx 2>/dev/null || echo ${TEST}B failed
sort -k2.3 -k2 in >xx &&
	sort -c -t: -k4n xx 2>/dev/null || echo ${TEST}C failed
sort -k2b,2.3 -k2 in >xx &&
	sort -c -t: -k5n xx 2>/dev/null || echo ${TEST}D failed
sort -k2.3,2.1b -k2 in >xx &&
	sort -c -t: -k6n xx 2>/dev/null || echo ${TEST}E failed
sort -k2,2.1b -k2r in >xx &&
	sort -c -t: -k7n xx 2>/dev/null || echo ${TEST}F failed
sort -b -k2,2 -k2 in >xx &&
	sort -c -t: -k8n xx 2>/dev/null || echo ${TEST}G failed
sort -b -k2,2b -k2 in >xx &&			# perhaps same as G
	sort -c -t: -k3n xx 2>/dev/null || echo ${TEST}H failed\
 "(standard is not clear on this)"

#---------------------------------------------------------------
TEST=26; echo $TEST	# empty fields, out of bounds fields
cat <<! >in
0 5
1 4
2 3
3 2
4 1
5 0
!
cp in out

xsort "" -k2.2,2.1 -k2.3,2.4

#---------------------------------------------------------------
TEST=27; echo $TEST	# displaced -o
rm -f out

$o sort /dev/null -o out || $o echo ${TEST}B failed
$o test -f out || $o echo ${TEST}C failed

#---------------------------------------------------------------
TEST=28; echo $TEST	# apparently nonmonotone field specs
cat <<! >in
aaaa c
x a
0 b
!
cp in out

$o xsort A +1 -0.3 +1.4 -1.5

xsort B -k2,1.3 -k2.5,2.5

#---------------------------------------------------------------
TEST=29; echo $TEST	# determination of end of option list
cat >-k <<!
x
!
rm -f out -c

sort -- -k </dev/null >xx || echo ${TEST}A argument failed
cmp xx -k || echo ${TEST}A comparison failed

sort - -c </dev/null 2>/dev/null && echo ${TEST}B failed

#---------------------------------------------------------------
TEST=30; echo $TEST	# missing newline
awk 'BEGIN{ printf "%s", "x"}' | sort >xx
wc -c <xx | awk '$1!=2{ print "'${TEST}' failed" }'

#---------------------------------------------------------------
TEST=31; echo $TEST	# -M, multiple fields
cat <<! >in
jan 10 1900
Feb 26 1900
feb 25 1900
January xx 1900
August 11 1900
jan 15 1990
feb 22 1990
mar 15 1990
apr 1 1990
may 45 1990
jun 14 1990
jul 4 1990
aug 1~ 1990
aug 11 1990
sep 1 1990
oct 12 1990
nov 24 1990
dec 25 1990
never 3 1990
 Dec 25 1990
!
cat <<! >out
January xx 1900
jan 10 1900
feb 25 1900
Feb 26 1900
August 11 1900
never 3 1990
jan 15 1990
feb 22 1990
mar 15 1990
apr 1 1990
may 45 1990
jun 14 1990
jul 4 1990
aug 1~ 1990
aug 11 1990
sep 1 1990
oct 12 1990
nov 24 1990
 Dec 25 1990
dec 25 1990
!

$M xsort "" -k3n -k1M -k2n

#---------------------------------------------------------------
TEST=32; echo $TEST	# -M case insensitivity, -r
cat <<! >in
x
june
january
december
!
cat <<! >out
december
june
january
x
!

$M xsort "" -Mr

#---------------------------------------------------------------
TEST=33; echo $TEST	# -g
cat <<! >in
2
1
10
.2
1e
1E1
1e.
!
cat <<! >out
.2
1
1e
1e.
2
10
1E1
!

$g xsort "" -g

#---------------------------------------------------------------
TEST=34; echo $TEST	# -g wide operands
cat <<! >in
.99999999999999999999
099999999999999999999e-21
099999999999999999999e-19
.1e1
!
cat <<! >out
099999999999999999999e-21
.99999999999999999999
.1e1
099999999999999999999e-19
!

$g xsort A -g

cat <<! >out
.1e1
.99999999999999999999
099999999999999999999e-19
099999999999999999999e-21
!

xsort B -n

#---------------------------------------------------------------
TEST=35; echo $TEST	#-g, -u with different fp reps
cat <<! >in
+0
-0
0.10
+.1
-.1
-100e-3
x
!
cat <<! >out
-.1
-100e-3
+0
-0
x
+.1
0.10
!

$g xsort A -g

$g sort -gu in >xx && $g sort -c -gu xx || echo ${TEST}B failed
$g linecount C xx 3

#---------------------------------------------------------------
TEST=36; echo $TEST	# -s
cat <<! >in
a 2
b 1
c 2
a 1
b 2
c 1
!
cat <<! >out
a 2
a 1
b 1
b 2
c 2
c 1
!

$s xsort "" -s -k1,1

#---------------------------------------------------------------
TEST=37; echo $TEST	# -s, multiple files
cat <<! >in
a 2
c 2
!
cat <<! >in1
a 1
b 1
c 1
!
cat <<! >out
c 2
b 1
a 2
!

$s sort -smru -k1,1 in in in1 in1 >xx
$s cmp xx out >/dev/null || echo $TEST failed

#---------------------------------------------------------------
TEST=38; echo $TEST	# -s
$s awk '
	BEGIN {
		for(i=1; i<50; i++)
			for(j=1; j<=i; j++) {
				print i, 2 >"in"
				print i, 1 >"in1"
			}
	}'

$s sort -m -s -k1,1n in in1 >out

$s awk '
	func stop()	{ print "'$TEST' failed"; exit }
	$1!=last1 	{ if(count!=last1 || $2!=2) stop();
			  count = 0}
	$1==last1 && $2!=last2 { if(count!=last1 || $2!=1) stop();
				 count = 0 }
			{ count++; last1 = $1; last2 = $2 }
	' out

#---------------------------------------------------------------
TEST=39; echo $TEST	# sort already sorted dictionary

dict=/usr/share/dict/words

sort -f $dict > out

cmp -s out $dict || echo $TEST failed - sort -f $dict

#---------------------------------------------------------------
TEST=40; echo "$TEST (long)"    # long lines test

line1="abcdefghijklmnopqrstabcdefghijklmnopqrstabcdefghijklmnopqrstabcdefghij"
line2="bcdefghijklmnopqrstabcdefghijklmnopqrstabcdefghijklmnopqrstabcdefghijk"

rm -f in out

# generate two out-of-order long lines (roughly 70 KB each), and try sort
# them
awk '
BEGIN {
	line1 = "'"${line1}"'"
	line2 = "'"${line2}"'"
	for(i=0; i<10; i++) {
		line1 = line1 line1
		line2 = line2 line2
	}
	print line2 > "in"
	print line1 > "in"

	print line1 > "out"
	print line2 > "out"
}'

xsort "A"

# generate already sorted file using $dict, only append space and
# long string to each line to make lines longer than usual
# use only first 200 lines (sort file approx. 14MB)
cat $dict | awk '
BEGIN {
	line = "'"$line1"'"
	for(i=0; i<10; i++)
		line = line line
	idx = 0
}
{ print $1 " " line; if (idx++ > 200) exit 0 }' > in
ln -sf in out
xsort "B" -f