awk/tests/T.misc

#!/bin/bash

# Copyright 2019 Joyent, Inc.

if [[ -z "$AWK" || -z "$WORKDIR" ]]; then
    printf '$AWK and $WORKDIR must be set\n' >&2
    exit 1
fi

TEMP0=$WORKDIR/test.temp.0
TEMP1=$WORKDIR/test.temp.1
TEMP2=$WORKDIR/test.temp.2
TEMP3=$WORKDIR/test.temp.3

RESULT=0

fail() {
	echo "$1" >&2
	RESULT=1
}

echo T.misc: miscellaneous buglets now watched for

rm -f core

echo 'The big brown over the lazy doe
The big brown over the lazy dog
x
The big brown over the lazy dog' > $TEMP0
echo 'failed
succeeded
failed
succeeded' > $TEMP1
$AWK '{ if (match($0, /^The big brown over the lazy dog/) == 0) {
		printf("failed\n")
	} else {
		printf("succeeded\n")
	}
} ' $TEMP0 > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc ghosh RE bug'

echo '123
1234567890
12345678901' > $TEMP0
echo '12345678901' > $TEMP1
$AWK 'length($0) > 10' $TEMP0 > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc last number bug'

# check some \ sequences in strings (ascii)
echo HIJKL > $TEMP1
echo $TEMP0 | $AWK '{ print "H\x49\x4a\x4BL" }' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc hex string cvt'

echo 012x45 > $TEMP1
$AWK 'BEGIN { print "0\061\62x\0645" }' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc oct string cvt'

# $i++ means ($i)++
echo 3 5 | $AWK '{ i = 1; print $i++ ; print $1, i }' > $TEMP1
echo '3
4 1' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc bad field increment'

# makes sure that fields are recomputed even if self-assignment
# take into account that subtracting from NF now rebuilds the record
echo 'a b c
s p q r
x y z' > $TEMP0
echo 'a
s p
x' > $TEMP1
$AWK '{ NF -= 2; $1 = $1; print }' < $TEMP0 > $TEMP2
diff $TEMP1 $TEMP2 || fail "BAD: T.misc bad field self-assignment"

echo '1
1' > $TEMP1
$AWK 'BEGIN {x = 1; print x; x = x; print x}' > $TEMP2
diff $TEMP1 $TEMP2 || fail "BAD: T.misc bad self-assignment"

echo 573109312 | $AWK '{print $1*4}' > $TEMP1
echo 2292437248 > $TEMP2
diff $TEMP1 $TEMP2 || fail "BAD: T.misc bad overflow"

# note that there are 8-bit characters in the echo
# some shells will probably screw this up.
echo '#
code € 1
code € 2' |
$AWK '/^#/' > $TEMP1
echo '#' > $TEMP2
diff $TEMP1 $TEMP2 || fail "BAD: T.misc bad match of 8-bit char"

echo hello |
$AWK 'BEGIN	{ FILENAME = "/etc/passwd" }
	{ print $0 }' >/dev/null
if [[ $? -eq 139 ]]; then fail "BAD: T.misc /etc/passwd dropped core"; fi

echo hello |
$AWK '  function foo(foo) {
                foo = 1
                foo()
        }
	{ foo(bar) }
' >/dev/null 2>&1
if [[ $? -eq 139 ]]; then
	fail "BAD: T.misc function foo(foo) dropped core"
	rm -f core
fi

echo '2
10' |
$AWK '{ x[NR] = $0 }	# test whether $0 is NUM as well as STR
END { if (x[1] > x[2]) print "BAD: T.misc: $0 is not NUM" }'


$AWK 'BEGIN {
	npad = substr("alexander" "           ",1,15)
	print npad
	}' > $TEMP0
grep '\\' $TEMP0 && fail "BAD: T.misc alexander fails"

# This should give an error about function arguments
$AWK '
function foo(x) { print "x is" x }
BEGIN { foo(foo) }
' 2> $TEMP0
grep "can't use function foo" $TEMP0 >/dev/null || fail "BAD: T.misc fcn args"


# gawk defref test; should give error about undefined function
$AWK 'BEGIN { foo() }' 2> $TEMP0
grep "calling undefined function foo" $TEMP0 >/dev/null || fail "BAD: T.misc undefined function"


# gawk arrayparm test; should give error about function
$AWK '
BEGIN {
    foo[1]=1;
    foo[2]=2;
    bug1(foo);
}
function bug1(i) {
    for (i in foo) {
	bug2(i);
	delete foo[i];
	print i,1,bot[1];
    }
}
function bug2(arg) {
    bot[arg]=arg;
}
' 2> $TEMP0
grep "can.t assign to foo" $TEMP0 >/dev/null || fail "BAD: T.misc foo bug"


# This should be a syntax error
$AWK '
!x = y
' 2> $TEMP0
grep "syntax error" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error !x=y fails"

# This should print bbb
$AWK '
BEGIN { up[1] = "a"
	for (i in up) gsub("a", "A", x)
	print x x "bbb"
	exit
      }
' > $TEMP0
grep bbb $TEMP0 >/dev/null || fail "BAD: T.misc gsub failed"

echo yes |
$AWK '
BEGIN {
	printf "push return" >"/dev/null"
	getline ans <"/dev/null"
} '
if [[ $? -eq 139 ]]; then fail "BAD: T.misc getline ans dropped core"; fi

$AWK 'BEGIN { unireghf() }
function unireghf(hfeed) { hfeed[1] = 0 }'
if [[ $? -eq 139 ]]; then fail "BAD: T.misc unireghf dropped core"; fi

echo x | $AWK '/[/]/' 2> $TEMP0
grep 'nonterminated character class' $TEMP0 >/dev/null || error 'BAD: T.misc nonterminated fails'
if [[ $? -eq 139 ]]; then fail "BAD: T.misc nonterminated dropped core"; fi

$AWK '
function f() { return 12345 }
BEGIN { printf "<%s>\n", f() }
' > $TEMP0
grep '<12345>' $TEMP0 >/dev/null || fail 'BAD: T.misc <12345> fails'

echo 'abc
def

ghi
jkl' > $TEMP0
$AWK '
BEGIN {	RS = ""
	while (getline <"'$TEMP0'")
		print
}' > $TEMP1
$AWK 'END {print NR}' $TEMP1 | grep 4 >/dev/null || fail 'BAD: T.misc abcdef fails'


# The following should not produce a warning about changing a constant
# nor about a curdled tempcell list
$AWK 'function f(x) { x = 2 }
BEGIN { f(1) }' > $TEMP0
grep '^' $TEMP0 && fail 'BAD: test constant change fails'

# The following should not produce a warning about a curdled tempcell list
$AWK 'function f(x) { x }
BEGIN { f(1) }' > $TEMP0
grep '^' $TEMP0 && fail 'BAD: test tempcell list fails'

$AWK 'BEGIN { print 9, a=10, 11; print a; exit }' > $TEMP1
echo '9 10 11
10' > $TEMP2
diff $TEMP1 $TEMP2 || fail 'BAD: T.misc (embedded expression)'

echo "abc defgh ijkl" | $AWK '
  { $1 = ""; line = $0; print line; print $0; $0 = line; print $0 }' > $TEMP1
echo " defgh ijkl
 defgh ijkl
 defgh ijkl" > $TEMP2
diff $TEMP1 $TEMP2 || fail 'BAD: T.misc (assignment to $0)'

$AWK '
function min(a, b)
{
	if (a < b)
		return a
	else
		return b
}
BEGIN { exit }
'
if [[ $? -eq 139 ]]; then fail "BAD: T.misc function min dropped core"; fi

# The following should not give a syntax error message:
$AWK '
function expand(chart) {
	getline chart < "CHAR.ticks"
}
' > $TEMP0
grep '^' $TEMP0 >/dev/null && fail 'BAD: T.misc expand error'

$AWK 'BEGIN { print 1e40 }' >/dev/null
if [[ $? -eq 139 ]]; then fail "BAD: T.misc 1E40 dropped core"; fi

# The following syntax error should not dump core:
$AWK '
$NF==3	{first=1}
$NF==2 && first==0 && (abs($1-o1)>120||abs($2-o2)>120)	{print $0}
$NF==2	{o1=%1; o2=$2; first=0}
' 2>/dev/null
if [[ $? -eq 139 ]]; then fail "BAD: T.misc first/abs dropped core"; fi

# The following syntax error should not dump core:
$AWK '{ n = split($1, address, !); print address[1] }' 2> $TEMP0
grep 'illegal statement' $TEMP0 >/dev/null || fail 'BAD: T.misc split error'
if [[ $? -eq 139 ]]; then fail "BAD: T.misc split! dropped core"; fi

# The following should cause a syntax error message
$AWK 'BEGIN {"hello"}' 2> $TEMP0
grep 'illegal statement' $TEMP0 >/dev/null || fail 'BAD: T.misc hello error'

# The following should give a syntax error message:
$AWK '
function pile(c,     r) {
	r = ++pile[c]
}

{ pile($1) }
' 2> $TEMP0
grep 'context is' $TEMP0 >/dev/null || fail 'BAD: T.misc pile error'

# This should complain about missing atan2 argument:
$AWK 'BEGIN { atan2(1) }' 2> $TEMP0
grep 'requires two arg' $TEMP0 >/dev/null || fail 'BAD: T.misc atan2 error'

# This should not core dump:
$AWK 'BEGIN { f() }
function f(A) { delete A[1] }
'
if [[ $? -eq 139 ]]; then fail "BAD: T.misc delete dropped core"; fi

# nasty one:  should not be able to overwrite constants
$AWK 'BEGIN { gsub(/ana/,"anda","banana")
		printf "the monkey ate a %s\n", "banana" }
' >/dev/null 2> $TEMP0
grep 'syntax error' $TEMP0 >/dev/null || fail 'BAD: T.misc gsub banana error'

# nasty one:  should not be able to overwrite constants
$AWK 'BEGIN { sub(/ana/,"anda","banana")
		printf "the monkey ate a %s\n", "banana" }
' >/dev/null 2> $TEMP0
grep 'syntax error' $TEMP0 >/dev/null || fail 'BAD: T.misc sub banana error'

# line numbers used to double-count comments
$AWK '#
#
#
/x
' >/dev/null 2> $TEMP0
grep 'line [45]' $TEMP0 >/dev/null || fail 'BAD: T.misc lineno'

echo 'x
\y' > $TEMP1
$AWK 'BEGIN { print "x\f\r\b\v\a\\y" }' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc weird chars'

echo 0 > $TEMP1
$AWK '	BEGIN { exit }
	{ print }
	END { print NR }' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc BEGIN exit'

echo 1 > $TEMP1
$AWK '	{ exit }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc immmediate exit'

echo 1 > $TEMP1
$AWK '	{i = 1; while (i <= NF) {if (i == NF) exit; i++ } }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc immmediate exit 2'

echo 1 > $TEMP1
$AWK '	function f() {
		i = 1; while (i <= NF) {if (i == NF) return NR; i++ }
	}
	{ if (f() == 1) exit }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc while return'

echo 1 > $TEMP1
$AWK '	function f() {
		split("a b c", arr)
		for (i in arr) {if (i == 3) return NR; i++ }
	}
	{ if (f() == 1) exit }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc while return'

echo 1 > $TEMP1
$AWK '	{i = 1; do { if (i == NF) exit; i++ } while (i <= NF) }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc immmediate exit 3'

echo 1 > $TEMP1
$AWK '	function f() {
		i = 1; do { if (i == NF) return NR; i++ } while (i <= NF)
	}
	{ if (f() == 1) exit }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc do return'

echo 1 > $TEMP1
$AWK '	{i = 1; do { if (i == NF) break; i++ } while (i <= NF); exit }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc immmediate exit 4'

echo 1 > $TEMP1
$AWK '	{ n = split($0, x)
	  for (i in x) {
		if (i == 1)
			exit } }
	END { print NR }' /etc/passwd > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc immmediate exit 5'

echo XXXXXXXX > $TEMP1
$AWK 'BEGIN { s = "ab\fc\rd\be"
	t = s;	gsub("[" s "]", "X", t); print t }' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc weird escapes in char class'

$AWK '{}' /etc/passwd glop/glop > $TEMP0 2> $TEMP2
grep "can't open.*glop" $TEMP2 >/dev/null || fail "BAD: T.misc can't open"

echo '


a
aa

b


c

' > $TEMP0
echo 3 > $TEMP1
$AWK 'BEGIN { RS = "" }; END { print NR }' $TEMP0 > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc RS botch'

$AWK 'BEGIN \
	{
		print "hello, world"
	}
}}}' > $TEMP1 2> $TEMP2
grep 'source line 5' $TEMP2 >/dev/null 2>&1 || fail 'BAD: T.misc continuation line number'


echo 111 222 333 > $TEMP0
$AWK '{ f[1]=1; f[2]=2; print $f[1], $f[1]++, $f[2], f[1], f[2] }' $TEMP0 > $TEMP2
echo 111 111 222 2 2 > $TEMP1
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc $f[1]++'


# These should be syntax errors
$AWK . 2> $TEMP0
grep "syntax error" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error . fails"

$AWK .. 2> $TEMP0
grep "syntax error" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error .. fails"

$AWK .E. 2> $TEMP0
grep "syntax error" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error .E. fails"

$AWK .++. 2> $TEMP0
grep "syntax error" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error .++. fails"


# These should be syntax errors
$AWK '$' 2> $TEMP0
grep "unexpected" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error $ fails"

$AWK '{print $' 2> $TEMP0
grep "unexpected" $TEMP0 >/dev/null || fail "BAD: T.misc syntax error \$2 fails"

$AWK '"' 2> $TEMP0
grep "non-terminated" $TEMP0 >/dev/null || fail "BAD: T.misc bare quote fails"


# %c of 0 is explicit null byte

echo '3' > $TEMP1
$AWK 'BEGIN {printf("%c%c\n", 0, 0) }' | wc | $AWK '{print $3}' > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc null byte'

# non-terminated RE

$AWK /xyz > $TEMP0 2>&1
grep "non-terminated" $TEMP0 >/dev/null || fail "BAD: T.misc non-terminated RE"

# next several were infinite loops, found by brian tsang.
# this is his example:

$AWK 'BEGIN {
    switch (substr("x",1,1)) {
    case /ask.com/:
	break
    case "google":
	break
    }
}' > $TEMP0 2>&1
grep "illegal statement" $TEMP0 >/dev/null || fail "BAD: T.misc looping syntax error 1"

$AWK 'BEGIN { s { c /./ } }' > $TEMP0 2>&1
grep "illegal statement" $TEMP0 >/dev/null || fail "BAD: T.misc looping syntax error 2"

$AWK 'BEGIN { s { c /../ } }' > $TEMP0 2>&1
grep "illegal statement" $TEMP0 >/dev/null || fail "BAD: T.misc looping syntax error 3"

$AWK 'BEGIN {printf "%2$s %1$s\n", "a", "b"}' >$TEMP0 2>&1
grep "'$' not permitted in awk formats" $TEMP0 >/dev/null || fail "BAD: T.misc '$' not permitted in formats"

echo 'a
b c
de fg hi' > $TEMP0
$AWK 'END { print NF, $0 }' $TEMP0 > $TEMP1
awk '{ print NF, $0 }' $TEMP0| tail -1 > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc END must preserve $0'

echo 'fg hi' > $TEMP0
$AWK 'END { print NF, $0 }' $TEMP0 > $TEMP1
awk '{ print NF, $0 }' $TEMP0| tail -1 > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc END must preserve $0'

echo '' > $TEMP0
$AWK 'END { print NF, $0 }' $TEMP0 > $TEMP1
awk '{ print NF, $0 }' $TEMP0| tail -1 > $TEMP2
cmp -s $TEMP1 $TEMP2 || fail 'BAD: T.misc END must preserve $0'


LC_ALL= LC_NUMERIC=ru_RU.ISO8859-5 $AWK 'BEGIN {
	"echo 1,200" | getline;
	if ($1 == 1.2) {
		printf "good ";
	} else {
		printf "bad ";
	}
	n = 2.3;
	print ($1 + 0.1), (n + 0.1);
}' > $TEMP1
echo 'good 1,3 2,4' > $TEMP2
diff $TEMP1 $TEMP2 || fail 'BAD: T.misc LC_NUMERIC should change radix'

$AWK 'function foo(q) {
	return (q = q);
}
BEGIN { print foo("h"); }' > $TEMP1
echo 'h' > $TEMP2
diff $TEMP1 $TEMP2 || fail 'BAD: T.misc return tempcell'

$AWK -v RECSIZE=8192 'BEGIN {
	for (c = 0; c < 3; c++) {
		a = (RECSIZE % 2 > 0 ? "5" : "55");
		while (length(a) < RECSIZE + c) {
			a = a " 5";
		}
		$0 = a;
		print $2;
	}
}' > $TEMP1
printf '5\n5\n5\n' > $TEMP2
diff $TEMP1 $TEMP2 || fail 'BAD: T.misc initial fields overflow'

exit $RESULT