1#!/usr/bin/env bash
2#
3# Subset of legacy regression tests: utf8
4#
5
6set -e
7
8if [ -z "$TESTBASE" ] ; then
9	echo TESTBASE needs to be passed via the env
10	exit 1
11fi
12
13. $TESTBASE/test-functions
14
15LC_ALL=en_US.UTF-8
16export LC_ALL
17
18DETOX=$1
19TABLEPATH=$(dirname $TESTBASE)/table
20METHOD1=utf_8
21METHOD2=safe
22METHOD3=wipeup
23
24# ---------------------------------------------------------------------------
25
26INPUT="safe string"
27OUTPUT="safe_string"
28
29test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
30
31# ---------------------------------------------------------------------------
32
33INPUT=$(printf "\\u00AE reg")
34OUTPUT="reg_reg"
35
36test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
37
38# ---------------------------------------------------------------------------
39
40INPUT=$(printf "\\u00A9 copy")
41OUTPUT="copy_copy"
42
43test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
44
45# ---------------------------------------------------------------------------
46
47INPUT=$(printf "\\u00C6 capital AE")
48OUTPUT="AE_capital_AE"
49
50test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
51
52# ---------------------------------------------------------------------------
53
54INPUT=$(printf "\\u00DE capital thorn")
55OUTPUT="TH_capital_thorn"
56
57test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
58
59# ---------------------------------------------------------------------------
60
61INPUT=$(printf "\\u014A capital ENG")
62OUTPUT="NG_capital_ENG"
63
64test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
65
66# ---------------------------------------------------------------------------
67
68INPUT=$(printf "\\u0172 capital U with Ogonek")
69OUTPUT="U_capital_U_with_Ogonek"
70
71test_sequence "$DETOX" "$INPUT" "$OUTPUT" "$TABLEPATH" "$METHOD1" "$METHOD2" "$METHOD3"
72
73