Windows: a test_cmp that is agnostic to random LF <> CRLF conversions

In a number of tests, output that was produced by a shell script is
compared to expected output using test_cmp. Unfortunately, the MSYS bash--
when invoked via git, such as in hooks--converts LF to CRLF on output
(as produced by echo and printf), which leads to many false positives.

Implements a diff tool that undoes the converted CRLF. To avoid that
sub-processes are spawned (which is very slow on Windows), the tool is
implemented as a shell function. Diff is invoked as usual only when a
difference is detected by the shell code.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

2 files changed, 67 insertions, 0 deletions

diff --git a/t/test-lib-functions.sh b/t/test-lib-functions.sh
index a7e9aacbb..f98769427 100644
--- a/t/test-lib-functions.sh
+++ b/t/test-lib-functions.sh
@@ -710,3 +710,69 @@ test_ln_s_add () {
 		git update-index --add --cacheinfo 120000 $ln_s_obj "$2"
 	fi
 }
+
+# The following mingw_* functions obey POSIX shell syntax, but are actually
+# bash scripts, and are meant to be used only with bash on Windows.
+
+# A test_cmp function that treats LF and CRLF equal and avoids to fork
+# diff when possible.
+mingw_test_cmp () {
+	# Read text into shell variables and compare them. If the results
+	# are different, use regular diff to report the difference.
+	local test_cmp_a= test_cmp_b=
+
+	# When text came from stdin (one argument is '-') we must feed it
+	# to diff.
+	local stdin_for_diff=
+
+	# Since it is difficult to detect the difference between an
+	# empty input file and a failure to read the files, we go straight
+	# to diff if one of the inputs is empty.
+	if test -s "$1" && test -s "$2"
+	then
+		# regular case: both files non-empty
+		mingw_read_file_strip_cr_ test_cmp_a <"$1"
+		mingw_read_file_strip_cr_ test_cmp_b <"$2"
+	elif test -s "$1" && test "$2" = -
+	then
+		# read 2nd file from stdin
+		mingw_read_file_strip_cr_ test_cmp_a <"$1"
+		mingw_read_file_strip_cr_ test_cmp_b
+		stdin_for_diff='<<<"$test_cmp_b"'
+	elif test "$1" = - && test -s "$2"
+	then
+		# read 1st file from stdin
+		mingw_read_file_strip_cr_ test_cmp_a
+		mingw_read_file_strip_cr_ test_cmp_b <"$2"
+		stdin_for_diff='<<<"$test_cmp_a"'
+	fi
+	test -n "$test_cmp_a" &&
+	test -n "$test_cmp_b" &&
+	test "$test_cmp_a" = "$test_cmp_b" ||
+	eval "diff -u \"\$@\" $stdin_for_diff"
+}
+
+# $1 is the name of the shell variable to fill in
+mingw_read_file_strip_cr_ () {
+	# Read line-wise using LF as the line separator
+	# and use IFS to strip CR.
+	local line
+	while :
+	do
+		if IFS=$'\r' read -r -d $'\n' line
+		then
+			# good
+			line=$line$'\n'
+		else
+			# we get here at EOF, but also if the last line
+			# was not terminated by LF; in the latter case,
+			# some text was read
+			if test -z "$line"
+			then
+				# EOF, really
+				break
+			fi
+		fi
+		eval "$1=\$$1\$line"
+	done
+}
diff --git a/t/test-lib.sh b/t/test-lib.sh
index 0fa7dfde7..77e487bd5 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -817,6 +817,7 @@ case $(uname -s) in
 	test_set_prereq NOT_CYGWIN
 	test_set_prereq SED_STRIPS_CR
 	test_set_prereq GREP_STRIPS_CR
+	GIT_TEST_CMP=mingw_test_cmp
 	;;
 *CYGWIN*)
 	test_set_prereq POSIXPERM