Merge branch 'unicode-tests' of https://github.com/dunhor/libarchive …

…into unicode-tests

Makefile.am

@@ -887,6 +887,7 @@ libarchive_test_EXTRA_DIST=\
 	libarchive/test/test_read_format_rar5_arm.rar.uu \
 	libarchive/test/test_read_format_rar5_blake2.rar.uu \
 	libarchive/test/test_read_format_rar5_compressed.rar.uu \
+	libarchive/test/test_read_format_rar5_unicode.rar.uu \
 	libarchive/test/test_read_format_rar5_different_window_size.rar.uu \
 	libarchive/test/test_read_format_rar5_different_solid_window_size.rar.uu \
 	libarchive/test/test_read_format_rar5_distance_overflow.rar.uu \

libarchive/test/test_read_format_rar5.c

@@ -24,6 +24,8 @@
  */
 #include "test.h"
 
+#include <locale.h>
+
 /* Some tests will want to calculate some CRC32's, and this header can
  * help. */
 #define __LIBARCHIVE_BUILD
@@ -806,36 +808,6 @@ DEFINE_TEST(test_read_format_rar5_extract_win32)
 	EPILOGUE();
 }
 
-DEFINE_TEST(test_read_format_rar5_unicode)
-{
-#if !defined(WIN32) || defined(__CYGWIN__)
-	skipping("Skipping test on non-Windows");
-	return;
-#else
-	/* Corresponds to the names:
-	 * 👋🌎.txt
-	 * 𝒮𝓎𝓂𝒷𝑜𝓁𝒾𝒸 𝐿𝒾𝓃𝓀.txt
-	 * Ⓗⓐⓡⓓ Ⓛⓘⓝⓚ.txt */
-	const wchar_t* emoji_name = L"\U0001f44b\U0001f30e.txt";
-	const wchar_t* italic_name = L"\U0001d4ae\U0001d4ce\U0001d4c2\U0001d4b7\U0001d45c\U0001d4c1\U0001d4be\U0001d4b8 \U0001d43f\U0001d4be\U0001d4c3\U0001d4c0.txt";
-	const wchar_t* circle_name = L"\u24bd\u24d0\u24e1\u24d3 \u24c1\u24d8\u24dd\u24da.txt";
-
-	PROLOGUE("test_read_format_rar5_unicode.rar");
-	assertA(0 == archive_read_next_header(a, &ae));
-	assertEqualWString(emoji_name, archive_entry_pathname_w(ae));
-	assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644);
-	assertA(0 == archive_read_next_header(a, &ae));
-	assertEqualWString(circle_name, archive_entry_pathname_w(ae));
-	assertEqualInt(archive_entry_mode(ae), AE_IFREG | 0644);
-	assertEqualWString(emoji_name, archive_entry_hardlink_w(ae));
-	assertA(0 == archive_read_next_header(a, &ae));
-	assertEqualWString(italic_name, archive_entry_pathname_w(ae));
-	assertEqualInt(archive_entry_mode(ae), AE_IFLNK | 0644);
-	assertEqualWString(emoji_name, archive_entry_symlink_w(ae));
-	EPILOGUE();
-#endif
-}
-
 DEFINE_TEST(test_read_format_rar5_block_by_block)
 {
 	/* This test uses strange buffer sizes intentionally. */
@@ -1406,3 +1378,99 @@ DEFINE_TEST(test_read_format_rar5_read_data_block_uninitialized_offset)
 
 	EPILOGUE();
 }
+
+DEFINE_TEST(test_read_format_rar5_unicode)
+{
+	PROLOGUE("test_read_format_rar5_unicode.rar");
+
+#if !defined(_WIN32) || defined(__CYGWIN__)
+	if (NULL == setlocale(LC_ALL, "en_US.UTF-8")) {
+		skipping("en_US.UTF-8 locale not available on this system.");
+		EPILOGUE();
+		return;
+	}
+#endif
+
+	/* Script "file.txt" */
+	assertA(0 == archive_read_next_header(a, &ae));
+	assertEqualInt(AE_IFREG, archive_entry_filetype(ae));
+#if defined(_WIN32) && !defined(__CYGWIN__)
+	assertEqualWString(L"\U0001d4bb\U0001d4be\U0001d4c1\U0001d452.txt",
+	    archive_entry_pathname_w(ae));
+#else
+	assertEqualString(
+	    "\xf0\x9d\x92\xbb\xf0\x9d\x92\xbe\xf0\x9d\x93\x81\xf0\x9d\x91\x92.txt",
+	    archive_entry_pathname(ae));
+#endif
+
+	/* Script "hardlink.txt" */
+	assertA(0 == archive_read_next_header(a, &ae));
+	assertEqualInt(AE_IFREG, archive_entry_filetype(ae));
+#if defined(_WIN32) && !defined(__CYGWIN__)
+	assertEqualWString(
+	    L"\U0001d4bd\U0001d4b6\U0001d4c7\U0001d4b9\U0001d4c1\U0001d4be\U0001d4c3\U0001d4c0.txt",
+	    archive_entry_pathname_w(ae));
+	assertEqualWString(L"\U0001d4bb\U0001d4be\U0001d4c1\U0001d452.txt",
+	    archive_entry_hardlink_w(ae));
+#else
+	assertEqualString(
+	    "\xf0\x9d\x92\xbd\xf0\x9d\x92\xb6\xf0\x9d\x93\x87\xf0\x9d\x92\xb9\xf0"
+	    "\x9d\x93\x81\xf0\x9d\x92\xbe\xf0\x9d\x93\x83\xf0\x9d\x93\x80.txt",
+	    archive_entry_pathname(ae));
+	assertEqualString(
+	    "\xf0\x9d\x92\xbb\xf0\x9d\x92\xbe\xf0\x9d\x93\x81\xf0\x9d\x91\x92.txt",
+	    archive_entry_hardlink(ae));
+#endif
+
+	/* Script "symlink.txt" */
+	assertA(0 == archive_read_next_header(a, &ae));
+	assertEqualInt(AE_IFLNK, archive_entry_filetype(ae));
+	assertEqualInt(AE_SYMLINK_TYPE_FILE, archive_entry_symlink_type(ae));
+#if defined(_WIN32) && !defined(__CYGWIN__)
+	assertEqualWString(
+	    L"\U0001d4c8\U0001d4ce\U0001d4c2\U0001d4c1\U0001d4be\U0001d4c3\U0001d4c0.txt",
+	    archive_entry_pathname_w(ae));
+	assertEqualWString(L"\U0001d4bb\U0001d4be\U0001d4c1\U0001d452.txt",
+	    archive_entry_symlink_w(ae));
+#else
+	assertEqualString("\xf0\x9d\x93\x88\xf0\x9d\x93\x8e\xf0\x9d\x93\x82\xf0"
+	    "\x9d\x93\x81\xf0\x9d\x92\xbe\xf0\x9d\x93\x83\xf0\x9d\x93\x80.txt",
+	    archive_entry_pathname(ae));
+	assertEqualString(
+	    "\xf0\x9d\x92\xbb\xf0\x9d\x92\xbe\xf0\x9d\x93\x81\xf0\x9d\x91\x92.txt",
+	    archive_entry_symlink(ae));
+#endif
+
+	/* Script "dir" */
+	assertA(0 == archive_read_next_header(a, &ae));
+	assertEqualInt(AE_IFDIR, archive_entry_filetype(ae));
+#if defined(_WIN32) && !defined(__CYGWIN__)
+	assertEqualWString(L"\U0001d4b9\U0001d4be\U0001d4c7",
+	    archive_entry_pathname_w(ae));
+#else
+	assertEqualString("\xf0\x9d\x92\xb9\xf0\x9d\x92\xbe\xf0\x9d\x93\x87",
+	    archive_entry_pathname(ae));
+#endif
+
+	/* Script "dir_symlink" */
+	assertA(0 == archive_read_next_header(a, &ae));
+	assertEqualInt(AE_IFLNK, archive_entry_filetype(ae));
+	assertEqualInt(AE_SYMLINK_TYPE_DIRECTORY, archive_entry_symlink_type(ae));
+#if defined(_WIN32) && !defined(__CYGWIN__)
+	assertEqualWString(
+	    L"\U0001d4b9\U0001d4be\U0001d4c7_\U0001d4c8\U0001d4ce\U0001d4c2\U0001d4c1\U0001d4be\U0001d4c3\U0001d4c0",
+	    archive_entry_pathname_w(ae));
+	assertEqualWString(L"\U0001d4b9\U0001d4be\U0001d4c7",
+	    archive_entry_symlink_w(ae));
+#else
+	assertEqualString("\xf0\x9d\x92\xb9\xf0\x9d\x92\xbe\xf0\x9d\x93\x87\x5f\xf0"
+	    "\x9d\x93\x88\xf0\x9d\x93\x8e\xf0\x9d\x93\x82\xf0\x9d\x93\x81\xf0\x9d"
+	    "\x92\xbe\xf0\x9d\x93\x83\xf0\x9d\x93\x80", archive_entry_pathname(ae));
+	assertEqualString("\xf0\x9d\x92\xb9\xf0\x9d\x92\xbe\xf0\x9d\x93\x87",
+	    archive_entry_symlink(ae));
+#endif
+
+	assertA(ARCHIVE_EOF == archive_read_next_header(a, &ae));
+
+	EPILOGUE();
+}