test_regexp

#! /usr/bin/env bash
SOURCE_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"

cd "${SOURCE_DIR}"

if [ "x$1" == "x" ]; then
	BIN="./regexp_dotno"
else
	BIN="$1"
fi

function err() { echo "$@" 1>&2; }


# Input has to be cut to be valid will return a cut (valid) version
function assert_equal_comp {
    IN="$1";
    EXPOUT="$2";
    GOTOUT="$(echo "$IN" | $BIN)";
    
    if [ "${EXPOUT}" != "${GOTOUT}" ]; then
        EXPLEN="$(echo -n "${EXPOUT}" | wc -c)"
        GOTLEN="$(echo -n "${GOTOUT}" | wc -c)"
        echo "[ERROR] Input '${IN}', expected '${EXPOUT}' (${EXPLEN}), got '${GOTOUT}' (${GOTLEN})";
    fi
}

# Input is a valid domain will return the same as input
function assert_equal {
    assert_equal_comp "$1" "$1";
}


# Input that can not be cut to a valid domain returns nothing
function assert_empty {
    IN="$1";
    GOTOUT="$(echo "$IN" | $BIN)";
    
    if [ "%" != "%${GOTOUT}" ]; then
        GOTLEN="$(echo -n "${GOTOUT}" | wc -c)"
        echo "[ERROR] Input '${IN}', expected '', got '${GOTOUT} (${GOTLEN})'";
    fi
}

# Check that maxlen is 63
assert_equal "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-exactly-62.no"
assert_equal "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-exactly-63.no"
assert_equal "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-exactly-63.oslo.no"
assert_equal_comp "wmaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-64-cut-to-63.no" "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-64-cut-to-63.no"
assert_equal_comp "wmaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-64-cut-to-63.oslo.no" "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-64-cut-to-63.oslo.no"
assert_equal_comp "wwwwwwwwwwwwwwwwmaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-80-cut-to-63.no" "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-80-cut-to-63.no"
assert_equal_comp "wwwwwwwwwwwwwwwwmaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-80-cut-to-63.oslo.no" "maaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-80-cut-to-63.oslo.no"

# Check that - is not permitted, either at start or end
assert_equal_comp "---foo.oslo.no" "foo.oslo.no"
assert_equal_comp "foo-.oslo.no" "oslo.no"
assert_equal_comp "-foo-.oslo.no" "oslo.no"
assert_empty "-o.no"
assert_empty "o-.no"
assert_empty "-o-.no"

# Check that domains can't be one char 
assert_empty "l.no"
assert_empty "-.no"
assert_empty "1.no"

# Check some legal hyphens
assert_equal "0--o.no"
assert_equal "0----------------o.no"
assert_equal "0-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o.no"

# Check that subdomains of category domains can be 1 char (oslo is category domain)
assert_equal "m.oslo.no"
assert_equal "mm.oslo.no"
assert_equal "mmm.oslo.no"
assert_equal_comp "det.er.m.oslo.no" "m.oslo.no" 
assert_equal_comp "m.m.m.oslo.no" "m.oslo.no"

# Check some weird category domains
assert_equal "www.gs.of.no"
assert_equal "foobar.sande.more-og-romsdal.no"
assert_equal "www.xn--hery-ira.xn--mre-og-romsdal-qqb.no"
assert_equal "våler.østfold.no"
assert_equal "sør-trøndelag.no"
assert_equal "ákŋoluokta.no"

# Test some numbers
assert_equal "1ákŋo2.no"
assert_equal "112.no"
assert_equal "99999999.no"

# Test some strings that will return multiple domains
assert_equal_comp "det.er.no.sol.no" "er.no
sol.no"
assert_equal_comp " det.er.no.sol.no" "er.no
sol.no"
assert_equal_comp "det.er.no.sol.no" "er.no
sol.no"
assert_equal_comp "det.er.no.oslo.no" "er.no
oslo.no"
assert_equal_comp "det.er.no.m.oslo.no" "er.no
m.oslo.no"
assert_equal_comp "oslo.no,sol.no" "oslo.no
sol.no"
assert_equal_comp "det.er.no.m.m.oslo.no.m.e.no" "er.no
m.oslo.no"  # oslo is category domain

# Test some domains with noise around them
assert_equal_comp "02inan.norsknettskole.no:80" "norsknettskole.no"
assert_equal_comp "02klho.norsknettskole.no/pages" "norsknettskole.no"
assert_equal_comp "02klho.norsknettskole.no/pages" "norsknettskole.no"

assert_equal "no.no"
assert_equal "øl.no"
assert_equal "lo.no"
assert_equal "lol.no"
assert_equal "o-a.no"
assert_equal "foobarfoooo.oslo.no"

assert_equal "02600.no"
assert_equal_comp "02bjli.norsknettskole.no" "norsknettskole.no"

assert_equal "0330.no"
assert_equal "123førstehjelp.no"
assert_equal_comp "02bat.norwegianbroker.no" "norwegianbroker.no"
assert_equal_comp "123ror.no123ferdig.no" "no123ferdig.no"
assert_equal_comp "123ror.rrno123ferdig.no" "rrno123ferdig.no"
assert_equal "1sol.no"
assert_equal "sol1.no"
assert_equal_comp "123ror.no123ferdig.no" "no123ferdig.no"
assert_equal "123áàäčçđéèêŋń-ñóòôöšŧüžæøå123.no"
assert_empty "1c.nov.ru"

echo "[INFO] All tests complete"