Add utility function to fixup Windows file paths

The idea is to iterate over a (canonicalized) path with forward
slashes by components, i.e. file and directory names. Each name
is then looked at by iterating over components, i.e. everything
between dots.

If a component is an illegal name, like COM1 or AUX, we add an
underscore. If it contains illegal characters, like : or \, we
re-map that character into unicode private use area.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

2 files changed, 59 insertions, 1 deletions

diff --git a/tests/libutil/Makemodule.am b/tests/libutil/Makemodule.am
index e039282..c783853 100644
--- a/tests/libutil/Makemodule.am
+++ b/tests/libutil/Makemodule.am
@@ -36,9 +36,12 @@ test_hex_decode_LDADD = libutil.a libcompat.a
 test_base64_decode_SOURCES = tests/libutil/base64_decode.c
 test_base64_decode_LDADD = libutil.a libcompat.a
 
+test_fix_win32_filename_SOURCES = tests/libutil/fix_win32_filename.c
+test_fix_win32_filename_LDADD = libutil.a libcompat.a
+
 LIBUTIL_TESTS = \
 	test_str_table test_rbtree test_xxhash test_threadpool test_ismemzero \
-	test_canonicalize_name test_filename_sane \
+	test_canonicalize_name test_filename_sane test_fix_win32_filename \
 	test_sdate_epoch test_hex_decode test_base64_decode
 
 check_PROGRAMS += $(LIBUTIL_TESTS)
diff --git a/tests/libutil/fix_win32_filename.c b/tests/libutil/fix_win32_filename.c
new file mode 100644
index 0000000..a4f71e8
--- /dev/null
+++ b/tests/libutil/fix_win32_filename.c
@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * fix_win32_filename.c
+ *
+ * Copyright (C) 2024 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "util/test.h"
+#include "util/util.h"
+
+static const struct {
+	const char *path;
+	const char *result;
+} test_data[] = {
+	{ "foo", "foo" },
+	{ "foo/bar", "foo/bar" },
+	{ "foo/bar.txt", "foo/bar.txt" },
+	{ "COM1", "COM1_" },
+	{ "COM1.txt", "COM1_.txt" },
+	{ "foo.aux", "foo.aux_" },
+	{ "foo/bar/test.LPT1/bla", "foo/bar/test.LPT1_/bla" },
+	{ "C:\\/foo/COM1.bla/bar",
+	  "C\xEF\x80\xBA\xEF\x81\x9c/foo/COM1_.bla/bar" },
+};
+
+int main(int argc, char **argv)
+{
+	(void)argc; (void)argv;
+
+	for (size_t i = 0; i < sizeof(test_data) / sizeof(test_data[0]); ++i) {
+		char *result = fix_win32_filename(test_data[i].path);
+		size_t out_len = strlen(test_data[i].result);
+
+		if (result == NULL) {
+			fprintf(stderr, "OOM for test case %u (%s)?\n",
+				(unsigned int)i, test_data[i].path);
+			return EXIT_FAILURE;
+		}
+
+		if (out_len != strlen(result) ||
+		    memcmp(result, test_data[i].result, out_len) != 0) {
+			fprintf(stderr,
+				"Mismatch for %s -> %s, got %s instead!\n",
+				test_data[i].path, test_data[i].result,
+				result);
+			free(result);
+			return EXIT_FAILURE;
+		}
+
+		free(result);
+	}
+
+	return EXIT_SUCCESS;
+}