t: refactor tests depending on Perl for textconv scripts

We have a couple of tests that depend on Perl for textconv scripts.
Refactor these tests to instead be implemented via shell utilities so
that we can drop a couple of PERL_TEST_HELPERS prerequisites.

Note that the conversion in t4030 is not a one-to-one equivalent to the
previous textconv script. Before this change we used to essentially do a
hexdump via Perl. The obvious conversion here would be to use `test-tool
hexdump` like we do for the other tests. But this would lead to a ripple
effect where we would have to adapt a bunch of other tests with a bunch
of seemingly unrelated changes, which would be somewhat awkward.

Instead, we're going with the minimum viable change: the test files we
write contain "\001" and "\000", and the test's expectation is that
those get translated into proper ASCII characters. So instead of doing a
full hexdump, we simply use tr(1) to translate these specific bytes.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

1 files changed, 3 insertions, 12 deletions

diff --git a/t/t4030-diff-textconv.sh b/t/t4030-diff-textconv.sh
index c7d8eb1245..f904fc19f6 100755
--- a/t/t4030-diff-textconv.sh
+++ b/t/t4030-diff-textconv.sh
@@ -4,12 +4,6 @@ test_description='diff.*.textconv tests'
 
 . ./test-lib.sh
 
-if ! test_have_prereq PERL_TEST_HELPERS
-then
-	skip_all='skipping diff textconv tests; Perl not available'
-	test_done
-fi
-
 find_diff() {
 	sed '1,/^index /d' | sed '/^-- $/,$d'
 }
@@ -26,13 +20,10 @@ cat >expect.text <<'EOF'
 +1
 EOF
 
-cat >hexdump <<'EOF'
-#!/bin/sh
-"$PERL_PATH" -e '$/ = undef; $_ = <>; s/./ord($&)/ge; print $_' < "$1"
-EOF
-chmod +x hexdump
-
 test_expect_success 'setup binary file with history' '
+	write_script hexdump <<-\EOF &&
+	tr "\000\001" "01" <"$1"
+	EOF
 	test_commit --printf one file "\\0\\n" &&
 	test_commit --printf --append two file "\\01\\n"
 '