clixon/test/test_search_index.sh
2023-02-15 13:59:21 +01:00

97 lines
2.7 KiB
Bash
Executable file

#!/usr/bin/env bash
# Test explicit search index.
# Test done by clixon explicit-index extension
# Test explicit indexes in lists these cases:
# - not a key string
# - not a key int
# - key in an ordered-by user
# - key in state data
# Use instance-id for tests, since api-path can only handle keys, and xpath is too complex.
# Magic line must be first in script (see README.md)
s="$_" ; . ./lib.sh || if [ "$s" = $0 ]; then exit 0; else return 0; fi
: ${clixon_util_path:=clixon_util_path -D $DBG -Y /usr/local/share/clixon}
# Number of list/leaf-list entries
: ${nr:=10000}
# Number of tests to generate XML for +1
max=2
# XML file (alt provide it in stdin after xpath)
for (( i=1; i<$max; i++ )); do
eval xml$i=$dir/xml$i.xml
done
ydir=$dir/yang
if [ ! -d $ydir ]; then
mkdir $ydir
fi
cat <<EOF > $ydir/moda.yang
module moda{
namespace "urn:example:a";
prefix a;
import clixon-config {
prefix "cc";
}
container x1{
description "extra index in list with single key";
list y{
ordered-by system;
key k1;
leaf k1{
type string;
}
leaf z{
type string;
}
leaf i{
description "explicit index variable";
type int32;
cc:search_index;
}
leaf j{
description "non-index variable";
type int32;
}
}
}
}
EOF
# Single string key
# Assign index i in reverse order
new "generate list with $nr single string key to $xml1"
echo -n '<x1 xmlns="urn:example:a">' > $xml1
for (( i=0; i<$nr; i++ )); do
let ii=$nr-$i-1
echo -n "<y><k1>a$i</k1><z>foo$i</z><i>$ii</i><j>$ii</j></y>" >> $xml1
done
echo -n '</x1>' >> $xml1
# First check correctness
for (( ii=0; ii<10; ii++ )); do
# key random
rnd=$(( ( RANDOM % $nr ) ))
# Let key index rndi be reverse of rnd
rndi=$(( $nr - $rnd - 1 ))
new "instance-id single string key i=$rndi (rnd:$rnd)"
expectpart "$($clixon_util_path -f $xml1 -y $ydir -p /a:x1/a:y[a:i=\"$rndi\"])" 0 "^0: <y><k1>a$rnd</k1><z>foo$rnd</z><i>$rndi</i><j>$rndi</j></y>$"
done
# Then measure time for index and non-index, assume correct
# For small nr, the time to parse is so much larger than searching (and also parsing involves
# searching) which makes it hard to make a test comparing accessing the index variable "i" and the
# non-index variable "j".
new "index search latency i=$rndi"
{ time -p $clixon_util_path -f $xml1 -y $ydir -p /a:x1/a:y[a:i=\"$rndi\"] -n 10 > /dev/null; } 2>&1 | awk '/real/ {print $2}'
new "non-index search latency j=$rndi"
{ time -p $clixon_util_path -f $xml1 -y $ydir -p /a:x1/a:y[a:j=\"$rndi\"] > /dev/null; } 2>&1 | awk '/real/ {print $2}'
rm -rf $dir
new "endtest"
endtest