WIP

1 files changed, 145 insertions, 0 deletions

diff --git a/utf/tst/test_utf32.cc b/utf/tst/test_utf32.cc
new file mode 100644
index 0000000..796b4cd
--- /dev/null
+++ b/utf/tst/test_utf32.cc
@@ -0,0 +1,145 @@
+#include "utf32.hh"
+
+#include "utf_error.hh"
+
+#include <gtest/gtest.h>
+
+TEST(utf32be, sanity) {
+  std::string_view str("\x00\x00\x00\x24", 4);
+  size_t offset = 0;
+  auto ret = utf::read32be(str, offset);
+  EXPECT_EQ('$', ret);
+  EXPECT_EQ(4, offset);
+
+  str = std::string_view("\x00\x00\x20\xAC", 4);
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  EXPECT_EQ(4, offset);
+
+  str = std::string_view("\x00\x01\x04\x37", 4);
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(0x10437, ret);
+  EXPECT_EQ(4, offset);
+}
+
+TEST(utf32le, sanity) {
+  std::string_view str("\x24\x00\x00\x00", 4);
+  size_t offset = 0;
+  auto ret = utf::read32le(str, offset);
+  EXPECT_EQ('$', ret);
+  EXPECT_EQ(4, offset);
+
+  str = std::string_view("\xAC\x20\x00\x00", 4);
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  EXPECT_EQ(4, offset);
+
+  str = std::string_view("\x37\x04\x01\x00", 4);
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(0x10437, ret);
+  EXPECT_EQ(4, offset);
+}
+
+TEST(utf32be, invalid) {
+  std::string_view str("\xFF\xFF\xFF\xFF");
+  size_t offset = 0;
+  auto ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00\x00\xD8\x00", 4);
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "";
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00", 1);
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00\x00", 2);
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00\x00\x00", 3);
+  offset = 0;
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+}
+
+TEST(utf32le, invalid) {
+  std::string_view str("\xFF\xFF\xFF\xFF");
+  size_t offset = 0;
+  auto ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00\xD8\x00\x00", 4);
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "";
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00", 1);
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00\x00", 2);
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = std::string_view("\x00\x00\x00", 3);
+  offset = 0;
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+}
+
+TEST(utf32be, bom) {
+  std::string_view str("\x00\x00\xFF\xFE\x00\x00\x20\xAC", 8);
+  size_t offset = 0;
+  auto ret = utf::read32be(str, offset);
+  EXPECT_EQ(0xFFFE, ret);
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  ret = utf::read32be(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(str.size(), offset);
+}
+
+TEST(utf32le, bom) {
+  std::string_view str("\xFE\xFF\x00\x00\xAC\x20\x00\x00", 8);
+  size_t offset = 0;
+  auto ret = utf::read32le(str, offset);
+  EXPECT_EQ(0xFFFE, ret);
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  ret = utf::read32le(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(str.size(), offset);
+}