diff options
Diffstat (limited to 'src/tests/manconv-odd-combinations')
| -rwxr-xr-x | src/tests/manconv-odd-combinations | 81 |
1 files changed, 81 insertions, 0 deletions
diff --git a/src/tests/manconv-odd-combinations b/src/tests/manconv-odd-combinations new file mode 100755 index 0000000..087d6fc --- /dev/null +++ b/src/tests/manconv-odd-combinations @@ -0,0 +1,81 @@ +#! /bin/sh + +# Test manconv's handling of various odd encoding combinations. + +: "${srcdir=.}" +# shellcheck source-path=SCRIPTDIR +. "$srcdir/testlib.sh" + +: "${MANCONV=manconv}" + +init + +(for x in $(seq 160 255); do + printf %b "\\$(printf %03o "$x")" +done +echo) >"$tmpdir/1.inp" + +iconv -f ISO-8859-1 -t UTF-8 <"$tmpdir/1.inp" >"$tmpdir/1.exp" +run $MANCONV -f UTF-8:ISO-8859-1 -t UTF-8 <"$tmpdir/1.inp" >"$tmpdir/1.out" +expect_files_equal '-f UTF-8:ISO-8859-1 -t UTF-8 on ISO-8859-1 input' \ + "$tmpdir/1.exp" "$tmpdir/1.out" + +iconv -f ISO-8859-2 -t UTF-8 <"$tmpdir/1.inp" >"$tmpdir/1-latin2.exp" +run $MANCONV -f UTF-8:ISO-8859-2 -t UTF-8 \ + <"$tmpdir/1.inp" >"$tmpdir/1-latin2.out" +expect_files_equal '-f UTF-8:ISO-8859-2 -t UTF-8 on ISO-8859-2 input' \ + "$tmpdir/1-latin2.exp" "$tmpdir/1-latin2.out" + +(for x in $(seq 1 1000); do + printf '‐' +done +echo 'Б' | iconv -f UTF-8 -t KOI8-R +echo '‐') >"$tmpdir/2.inp" +iconv -f KOI8-R -t UTF-8 <"$tmpdir/2.inp" >"$tmpdir/2.exp" +run $MANCONV -f UTF-8:KOI8-R -t UTF-8 <"$tmpdir/2.inp" >"$tmpdir/2.out" +expect_files_equal \ + '-f UTF-8:KOI8-R -t UTF-8 on KOI8-R input with UTF-8 prefix' \ + "$tmpdir/2.exp" "$tmpdir/2.out" + +(for x in $(seq 160 255); do + printf %b "\\$(printf %03o "$x")" +done +echo) | iconv -f ISO-8859-1 -t UTF-8 >"$tmpdir/3.inp" +run $MANCONV -f UTF-8:ISO-8859-1 -t UTF-8 <"$tmpdir/3.inp" >"$tmpdir/3.out" +expect_files_equal '-f UTF-8:ISO-8859-1 -t UTF-8 preserves UTF-8 input' \ + "$tmpdir/3.inp" "$tmpdir/3.out" + +# U+00B7 MIDDLE DOT is not representable in ISO-8859-2, and so should be +# omitted. However, manconv should still recognise that the input was UTF-8 +# rather than falling back to ISO-8859-2. +cat >"$tmpdir/4.inp" <<'EOF' +š·ł +EOF +iconv -f UTF-8 -t ISO-8859-2 >"$tmpdir/4.exp" <<EOF +šł +EOF +run $MANCONV -f UTF-8:ISO-8859-2 -t ISO-8859-2//IGNORE \ + <"$tmpdir/4.inp" >"$tmpdir/4.out" +expect_files_equal \ + 'recognises input encoding and omits invalid output character' \ + "$tmpdir/4.exp" "$tmpdir/4.out" + +# 0xAE does not exist in ISO-8859-7, so manconv won't be able to recode this +# to UTF-8 without conversion errors. (In the original case where this was +# seen in the wild, the coding: tag should actually have read ISO-8859-13.) +iconv -f UTF-8 -t ISO-8859-13 >"$tmpdir/5.inp" <<'EOF' +'\" -*- coding: ISO-8859-7 +REGISTERED SIGN: ® +trailing data +EOF +cat >"$tmpdir/5.exp" <<'EOF' +'\" -*- coding: UTF-8 +EOF +<"$tmpdir/5.inp" tail -n +2 | iconv -f ISO-8859-7 -t UTF-8//IGNORE \ + >>"$tmpdir/5.exp" 2>/dev/null +run $MANCONV -f UTF-8:ISO-8859-1 -t UTF-8//IGNORE \ + <"$tmpdir/5.inp" >"$tmpdir/5.out" +expect_files_equal 'copes with invalid input characters' \ + "$tmpdir/5.exp" "$tmpdir/5.out" + +finish |