WIP

1 files changed, 157 insertions, 0 deletions

diff --git a/utf/tst/test_utf16.cc b/utf/tst/test_utf16.cc
new file mode 100644
index 0000000..c17982e
--- /dev/null
+++ b/utf/tst/test_utf16.cc
@@ -0,0 +1,157 @@
+#include "utf16.hh"
+
+#include "utf_error.hh"
+
+#include <gtest/gtest.h>
+
+TEST(utf16be, sanity) {
+  std::string_view str("\x00\x24", 2);
+  size_t offset = 0;
+  auto ret = utf::read16be(str, offset);
+  EXPECT_EQ('$', ret);
+  EXPECT_EQ(2, offset);
+
+  str = "\x20\xAC";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  EXPECT_EQ(2, offset);
+
+  str = "\xD8\x01\xDC\x37";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(0x10437, ret);
+  EXPECT_EQ(4, offset);
+
+  str = "\xD8\x52\xDF\x62";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(0x24B62, ret);
+  EXPECT_EQ(4, offset);
+}
+
+TEST(utf16le, sanity) {
+  std::string_view str("\x24\x00", 2);
+  size_t offset = 0;
+  auto ret = utf::read16le(str, offset);
+  EXPECT_EQ('$', ret);
+  EXPECT_EQ(2, offset);
+
+  str = "\xAC\x20";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  EXPECT_EQ(2, offset);
+
+  str = "\x01\xD8\x37\xDC";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(0x10437, ret);
+  EXPECT_EQ(4, offset);
+
+  str = "\x52\xD8\x62\xDF";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(0x24B62, ret);
+  EXPECT_EQ(4, offset);
+}
+
+TEST(utf16be, bom) {
+  std::string_view str("\xFE\xFF\x20\xAC");
+  size_t offset = 0;
+  auto ret = utf::read16be(str, offset);
+  EXPECT_EQ(0xFEFF, ret);
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(str.size(), offset);
+}
+
+TEST(utf16le, bom) {
+  std::string_view str("\xFF\xFE\xAC\x20");
+  size_t offset = 0;
+  auto ret = utf::read16le(str, offset);
+  EXPECT_EQ(0xFEFF, ret);
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(str.size(), offset);
+}
+
+TEST(utf16be, invalid) {
+  std::string_view str("\xD8");
+  size_t offset = 0;
+  auto ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xD8\x01";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xD8\x01\xDC";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xDC\x37\xD8\x01";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xD8\x01\xD8\x01";
+  offset = 0;
+  ret = utf::read16be(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+}
+
+TEST(utf16le, invalid) {
+  std::string_view str("\x01");
+  size_t offset = 0;
+  auto ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\x01\xD8";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\x01\xD8\x37";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\x37\xDC\x01\xD8";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\x01\xD8\x01\xD8";
+  offset = 0;
+  ret = utf::read16le(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+}