WIP

1 files changed, 188 insertions, 0 deletions

diff --git a/utf/tst/test_utf8.cc b/utf/tst/test_utf8.cc
new file mode 100644
index 0000000..10df969
--- /dev/null
+++ b/utf/tst/test_utf8.cc
@@ -0,0 +1,188 @@
+#include "utf8.hh"
+
+#include "utf_error.hh"
+
+#include <gtest/gtest.h>
+
+TEST(utf8, sanity) {
+  std::string_view str("$");
+  size_t offset = 0;
+  auto ret = utf::read8(str, offset);
+  EXPECT_EQ('$', ret);
+  EXPECT_EQ(1, offset);
+
+  str = "\xC2\xA3";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0xa3, ret);
+  EXPECT_EQ(2, offset);
+
+  str = "\xD0\x98";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x418, ret);
+  EXPECT_EQ(2, offset);
+
+  str = "\xE0\xA4\xB9";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x939, ret);
+  EXPECT_EQ(3, offset);
+
+  str = "\xE2\x82\xAC";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x20AC, ret);
+  EXPECT_EQ(3, offset);
+
+  str = "\xED\x95\x9C";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0xD55C, ret);
+  EXPECT_EQ(3, offset);
+
+  str = "\xF0\x90\x8D\x88";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x10348, ret);
+  EXPECT_EQ(4, offset);
+}
+
+TEST(utf8, overlong) {
+  std::string_view str("\xF0\x82\x82\xAC");
+  size_t offset = 0;
+  auto ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xE0\x81\x81";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xC0\x80";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+}
+
+TEST(utf8, invalid) {
+  std::string_view str("\xED\xB0\x80");
+  size_t offset = 0;
+  auto ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xFB\xFF\xFF";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xFF\xFF\xFF\xFF\xFF";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\x80";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xC2";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xC2\x03";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::INVALID, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xE0\xA4";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+
+  str = "\xF0\x90\x8D";
+  offset = 0;
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(0, offset);
+}
+
+TEST(utf8, multiple1) {
+  std::string_view str("\x4D\xC3\xAC\x6E\x68\x20\x6E\xC3\xB3\x69\x20\x74\x69"
+                       "\xE1\xBA\xBF\x6E\x67\x20\x56\x69\xE1\xBB\x87\x74");
+  size_t offset = 0;
+  auto ret = utf::read8(str, offset);
+  EXPECT_EQ('M', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0xEC, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('n', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('h', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(' ', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('n', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0xF3, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('i', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(' ', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('t', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('i', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x1EBF, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('n', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('g', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(' ', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('V', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('i', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x1EC7, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ('t', ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(str.size(), offset);
+}
+
+TEST(utf8, multiple2) {
+  std::string_view str("\xF0\xA8\x89\x9F\xE5\x91\x90\xE3\x97\x82\xE8\xB6\x8A");
+  size_t offset = 0;
+  auto ret = utf::read8(str, offset);
+  EXPECT_EQ(0x2825F, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x5450, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x35C2, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(0x8D8A, ret);
+  ret = utf::read8(str, offset);
+  EXPECT_EQ(utf::NEED_MORE, ret);
+  EXPECT_EQ(str.size(), offset);
+}