#!/usr/bin/env ksh
# $Id: ouereg.sh,v 1.8 2009/10/10 20:05:46 ksb Exp $
# regression test for oue, never said it'd be easy.

: ${oue_path:=oue}

: ${TMPDIR:=/tmp}
while T1=$TMPDIR/oAt$$,e$((RANDOM%1000)) ; do
	[ -f $T1 ] || break
done
while T2=$TMPDIR/oAt$$,o$((RANDOM%1000)) ; do
	[ -f $T2 ] || break
done
D1=$T1.db D2=$T2.db
export T1 T2 D1 D2
trap 'rm -f $T1 $T2 $D1 $D2' EXIT
cat <<\! >$T1
a
b
!
cat <<\! >$T2
b
c
c
!

# overhead checks, all tools I code must honor these.
$oue_path -V >/dev/null || exit 10
$oue_path -h >/dev/null || exit 11
$oue_path -H >/dev/null || exit 12

# trivial case tests, with -span and -k
[ _one = _"`(echo one) |$oue_path`" ] || exit 20
[ _one = _"`(echo one; echo one; echo one; echo one) |$oue_path`" ] || exit 20
[ _"one
two" = _"`(echo one; echo two; echo one; echo two) |$oue_path`" ] || exit 20
[ _"one
two" = _"`(echo one; echo two; echo one; echo two) |$oue_path -2`" ] || exit 20
[ _"one" = _"`(echo one; echo two; echo one; echo two) |$oue_path -2 -k %1`" ] || exit 20
[ _"two" = _"`(echo one; echo two; echo one; echo two) |$oue_path -2 -k %2`" ] || exit 20
[ 2 -eq `$oue_path $T1 |wc -l` ] || exit 22
[ 2 -eq `$oue_path $T2 |wc -l` ] || exit 22
[ 3 -eq `$oue_path $T1 $T2 |wc -l` ] || exit 22

#a c      change the escape character, default '%', try '+'
# %% and %0
[ _"%1" = _"`echo 1 |$oue_path -a+ -R %+1`" ] || exit 25
[ _"%1" = _"`echo %1 |$oue_path -a+ -R +1`" ] || exit 25
[ _"@1" = _"`echo @1 |$oue_path -a+`" ] || exit 26
[ _"@1
@3" = _"`(echo @1 ; echo @3) |$oue_path -2 -a+`" ] || exit 27
[ _"+2" = _"`echo 2 |$oue_path -a+ -R +++1`" ] || exit 28
[ _"" = _"`echo 2 |$oue_path -a+ -R +0`" ] || exit 29

#c count unique keys and -s
[ -z "`$oue_path -c $T1| egrep -v '^1 [ab]$'`" ] || exit 80
[ -z "`$oue_path -c $T2| egrep -v '^(1 b|2 c)$'`" ] || exit 80
[ -z "`$oue_path -c $T1 $T2| egrep -v '^(2 [bc]|1 a)$'`" ] || exit 80
[ _"1 a 2 b 2 c "_ = _"`$oue_path -cs $T1 $T2 |tr -s ' \n' '  '`"_ ] || exit 80
#c with -v doesn't change anything -- hard to test that it doesn't

#d output only duplicate lines (-c, -v)
[ -z "`$oue_path -d $T1`" ] || exit 81
[ -z "`$oue_path -d $T2| grep -v '^c$'`" ] || exit 81
[ -z "`$oue_path -d $T1 $T2| egrep -v '^[bc]$'`" ] || exit 81
[ _"a" = _"`$oue_path -d $T1 $T1 |egrep -v '^[b]$'`" ] || exit 81
#d with -v or -s
[ _"b c "_ = _"`$oue_path -ds $T1 $T2 |tr -s ' \n' '  '`"_ ] || exit 82
[ _"b c "_ = _"`$oue_path -dls $T2 $T1 |tr -s ' \n' '  '`"_ ] || exit 82
[ _"a" = _"`$oue_path -dv $T1 $T2`" ] || exit 82
[ _"a" = _"`$oue_path -dlv $T1 $T2`" ] || exit 82

#D state  save key values in state dbm between runs
rm -f $D1 $D2
$oue_path -D $D1 /dev/null
[ -f $D1 ] || exit 30
[ 2 -eq `$oue_path -D $D1 $T1 |wc -l` ] || exit 32
[ 2 -eq `$oue_path -D $D2 $T2 |wc -l` ] || exit 32
[ 1 -eq `$oue_path -D $D1 $T1 $T2 |wc -l` ] || exit 33
[ 1 -eq `$oue_path -D $D2 $T1 $T2 |wc -l` ] || exit 33
[ 0 -eq `$oue_path -D $D1 $T1 $T2 |wc -l` ] || exit 33

#c with -D, -I
rm -f $D1 $D2
[ 2 -eq "`$oue_path -D $D1 -c $T1 |wc -l`" ] || exit 83
[ 0 -eq "`$oue_path -I $D1 -c $T1 |grep -v '^2' |wc -l`" ] || exit 83
[ 0 -eq "`$oue_path -I $D1 -c $T1 $T1 |grep -v '^3' |wc -l`" ] || exit 83

#d with -D, -I, and -v
# t1 has no dups, t2 has "c" in it,   t1+t2 has b and c as dups
rm -f $D1 $D2
[ 0 -eq "`$oue_path -D $D1 -d $T1 |wc -l`" ] || exit 84
[ _c = _"`$oue_path -D $D1 -d $T2`" ] || exit 84
[ 2 -eq "`$oue_path -I $D1 -d $T1 $T2 |wc -l`" ] || exit 84
[ 1 -eq "`$oue_path -I $D1 -dv $T1 $T2 |wc -l`" ] || exit 84

#d with -c, and -v
rm -f $D1 $D2
[ _"2 c" = _"`$oue_path -D $D1 -dc $T2`" ] || exit 85
[ _"1 b" = _"`$oue_path -D $D1 -vdc $T2`" ] || exit 85
[ 2 -eq "`$oue_path -I $D1 -i /dev/null| wc -l`" ] || exit 85


#I prev   consult the contents of a previous state dbm first
#i        include previous values at startup
rm -f $D1 $D2
echo z |$oue_path -D $D1 -R ''
[ -f $D1 ] || exit 30
# see we don't update prev
[ 2 -eq `$oue_path -I $D1 $T1 |wc -l` ] || exit 35
[ 2 -eq `$oue_path -I $D1 $T2 |wc -l` ] || exit 35
[ 1 -eq `$oue_path -I $D1 -i /dev/null |wc -l` ] || exit 35

#x and -B
rm -f $D1 $D2
$oue_path -cD $D1 -r "%f:%n:%t" $T2 >/dev/null	# 1 b, 2 c
[ -z "`$oue_path -iI $D1 -B "%v from %k" /dev/null |egrep -v '^[^:]*:([0-9]*):\1 from [abc]$'`" ] || exit 86
# 1 b + 1 in t1 outputs 2 b, inverse is 1 a
[ _"2 b" = _"`$oue_path -dc -I $D1 -x '%[v:3]' $T1`" ] || exit 86
[ _"1 a" = _"`$oue_path -dcv -I $D1 -x '%[v:3]' $T1`" ] || exit 86

#v        invert the filter to only include elements from prev
# also to see that we consult prev, add $T2 (c and b) to the z in it, count them
rm -f $D1 $D2
$oue_path -D $D1 -R '' $T2
[ _'. . '_ = "_`$oue_path -I $D1 -iR. /dev/null| tr -s '\n' ' '`_" ] || exit 36
# only 'a' is unique below (from $T1) 'b', 'c', and 'z' are in $D1
[ 1 -eq `$oue_path -I $D1 $T1 |wc -l` ] || exit 37		# a
[ 0 -eq `$oue_path -I $D1 $T2 |wc -l` ] || exit 37		# <empty>
[ 1 -eq `$oue_path -I $D1 $T1 $T2 |wc -l` ] || exit 37		# a
[ 1 -eq `$oue_path -vI $D1 $T1 |wc -l` ] || exit 38		# b
[ 2 -eq `$oue_path -vI $D1 $T2 |wc -l` ] || exit 39		# b c
[ 2 -eq `$oue_path -vI $D1 $T1 $T2 |wc -l` ] || exit 39		# b c
rm -f $D1 $D2

#R report report on key/value pairs in each dbm, word and the empty string
[ 0 -eq `$oue_path -R a $T1 |grep plus |wc -l` ] || exit 26
[ 2 -eq `$oue_path -R plus $T1 |grep plus |wc -l` ] || exit 26
[ 0 -eq `$oue_path -R '' $T1 |wc -l` ] || exit 27

#r memory dicer value-map for each key in our state file, both %m and %v
# should be the previous memory under -I ... -i
$oue_path -D $D1 -r oO -R '' $T2
[ _'oO oO '_ = "_`$oue_path -I $D1 -iR "%m" /dev/null| tr -s '\n' ' '`_" ] || exit 40
[ _'oO oO '_ = "_`$oue_path -I $D1 -iR "%v" /dev/null| tr -s '\n' ' '`_" ] || exit 41
# under -vI we get more action.
[ _'oO nN oO nN '_ = "_`$oue_path -vI $D1 -r "nN" -R "%v %m" $D1 $T2| tr -s '\n' ' '`_" ] || exit 42
rm -f $D1 $D2

#z        read find-print0 output as input files
[ _"thing!", = "_`find $T1 $T2 -print0 |oue -z -2 -R thing |tr '\000' '!'`," ] || exit 50
#
#p pad    complete short records with this token
# %*     The element's lines joined with spaces.
# %@     The element's lines joined with newlines (viz. '\n').
[ _"$T1 miss!", = "_`find $T1 -print0 |oue -p miss -z -2 -R '%*' |tr '\000' '!'`," ] || exit 50
[ _"$T1
miss!", = "_`find $T1 -print0 |oue -p miss -z -2 -R '%@' |tr '\000' '!'`," ] || exit 51

#%n     The line number from that file.
[ _'1.c' = _`$oue_path -3 -k "%n.%3" $T2` ] || exit 54
#%$     The last line in the element, as in xapply.
[ _'1.c' = _`$oue_path -3 -k '%n.%\$' $T2` ] || exit 55
#%f     The file which provided the element.  The expander %i represents
#the position of that file on the command-line (1, 2, 3, ... N).
[ _"$T2,1" = _`$oue_path -3 -k "%f,%i" $T2` ] || exit 56
[ _'1.a
1.b
2.c'_ = _"`$oue_path -2 -k '%n.%1' $T1 $T2`_" ] || exit 57

#%u     The count of the unique elements discovered so far.
[ _"b,0,c
c,1,ksb" = _"`$oue_path -2 -p ksb -k "%1,%u,%2" $T2`" ] || exit 58

# Check counting markup
rm -f $D1 $D2
[ _"1.b 2.c " = _"`$oue_path -c -D $D1 -r "%c" -R "%m.%k" $T2 |sort -n |tr -s '\n ' '  '`" ] || exit 58
[ _"1.a 1.b " = _"`$oue_path -c -I $D1 -r "%c" -R "%m.%k" -x "%v" $T1 |sort -n |tr -s '\n ' '  '`" ] || exit 58
[ _"1.a 2.b " = _"`$oue_path -c -I $D1 -r "%t" -R "%m.%k" -x "%v" $T1 |sort -n |tr -s '\n ' '  '`" ] || exit 58
[ _"0.a 1.b " = _"`$oue_path -c -I $D1 -r "%o" -R "%m.%k" -x "%v" $T1 |sort -n |tr -s '\n ' '  '`" ] || exit 58
[ _"0.a 10.b " = _"`$oue_path -c -I $D1 -r "%o" -R "%m.%k" -x "%v0" $T1 |sort -n |tr -s '\n ' '  '`" ] || exit 58
[ _"1.b " = _"`$oue_path -d -I $D1 -r "%o" -R "%m.%k" -x "%[v.1]" $T1 |sort -n |tr -s '\n ' '  '`" ] || exit 58
# under -d we have to cross 2, not jump over it (10+1 != 2) -- ksb
[ _"" = _"`$oue_path -d -I $D1 -r "%o" -R "%m.%k" -x "%[v.1]0" $T1 |sort -n |tr -s '\n ' '  '`" ] || exit 58
rm -f $D1 $D2

#e every update
#f first update
[ _2.c = _"`$oue_path -e "%n" -R "%e.%k" $T2 |grep c`" ] || exit 60
[ _2.c = _"`$oue_path -f "%n" -R "%e.%k" $T2 |grep c`" ] || exit 60
[  _@.c = _"`$oue_path -f "@"          -R "%e.%k" $T2 |grep c`" ] || exit 60
[ _!@.c = _"`$oue_path -f "@" -e "!%p" -R "%e.%k" $T2 |grep c`" ] || exit 60
[ _!@.c = _"`echo c |$oue_path -f "@" -e "!%p" -R "%e.%k" - $T2 |grep c`" ] || exit 60

#l last element instance (-c always inplies -l)
[ _!!!@.c = _"`echo c |$oue_path -lf "@" -e "!%p" -R "%e.%k" - $T2 |grep c`" ] || exit 61
[ _2.c = _"`$oue_path -d -R '%n.%k' $T2`" ] || exit 61
[ _3.c = _"`$oue_path -ld -R '%n.%k' $T2`" ] || exit 61
[   _-:1.c = _"`echo c |$oue_path     -r '%f:%n.%k' -R "%m" - $T2 |grep c`" ] || exit 62
[ _$T2:3.c = _"`echo c |$oue_path  -l -r '%f:%n.%k' -R "%m" - $T2 |grep c`" ] || exit 62
[   _-:1.c = _"`echo c |$oue_path  -d -r '%f:%n.%k' -R "%m" - $T2`" ] || exit 62
[ _$T2:2.c = _"`echo c |$oue_path  -d -r '%f:%n.%k' -R "%m" $T2 -`" ] || exit 62
[ _$T2:3.c = _"`echo c |$oue_path -ld -r '%f:%n.%k' -R "%m" - $T2`" ] || exit 63
[ _-:1.c = _"`echo c |$oue_path -ld -r '%f:%n.%k' -R "%m" $T2 -`" ] || exit 63
[ _3.c = _"`$oue_path -le "%n" -R "%e.%k" $T2 |grep c`" ] || exit 59
[ _2.c = _"`$oue_path -lf "%n" -R "%e.%k" $T2 |grep c`" ] || exit 59
[ _!!@.c = _"`$oue_path -l -f "@" -e "!%p" -R "%e.%k" $T2 |grep c`" ] || exit 59

#s stable order (under -c, -d, -l)
[ _"1 a 2 b 2 c " = _"`$oue_path -cs $T1 $T2 |tr -s '\n ' '  '`" ] || exit 18
[ _"b c " = _"`$oue_path -ds $T1 $T2 |tr -s '\n ' '  '`" ] || exit 18
[ _"b c " = _"`echo c |$oue_path -d  - $T1 $T2 |tr -s '\n ' '  '`" ] || exit 19
[ _"c b " = _"`echo c |$oue_path -ds - $T1 $T2 |tr -s '\n ' '  '`" ] || exit 19

#b length specify a bigger dicer buffer size (default "10k", try 45k + 1block)
# (note that the code rounds up, so you can't get an unaligned or odd length)
[ 10240 = `$oue_path -V |sed -n -e 's/.*buffer length \([0-9]*\)$/\1/p'` ] || exit 34
[ 46592 = `$oue_path -b 45k+1b -V |sed -n -e 's/.*buffer length \([0-9]*\)$/\1/p'` ] || exit 34

# not tested (yet or ever)
#N        never lock either dbm file for updates (I don't know how)

# check for the embeded NUL bug in 2.17
[ _zero! = _"`echo 'zero' |oue |tr '\000' '!'`" ] && exit 101

exit 0