// unicode_test.cc - Don Yang (uguu.org)
//
// 06/22/11

#include"unicode.h"
#include"util.h"

namespace
{

// Wrapper to ParseUtf8Char, calls the function with a string, and
// verifies that it's fully parsed and returns the expected value.
// Returns true on success.
static bool CheckParseUtf8Char(const char *input, unsigned int expected)
{
   const string input_str(input);
   string::const_iterator i = input_str.begin();
   const unsigned int actual = ParseUtf8Char(input_str, &i);
   if( expected != actual )
   {
      printf("ERROR: decoded values mismatched: %04X vs %04X",
             expected, actual);
      return false;
   }
   if( expected != 0xfffd )
   {
      ++i;
      if( i != input_str.end() )
      {
         puts("ERROR: string not fully parsed");
         return false;
      }
   }
   return true;
}

// Test UTF-8 parsing
static void TestParseUtf8()
{
   CHECK(CheckParseUtf8Char("\xd0\x80", 1 << 10), "2 bytes");
   CHECK(CheckParseUtf8Char("\xe8\x80\x80", 1 << 15), "3 bytes");
   CHECK(CheckParseUtf8Char("\xf4\x80\x80\x80", 1 << 20), "4 bytes");
   CHECK(CheckParseUtf8Char("\xfa\x80\x80\x80\x80", 1 << 25), "5 bytes");
   CHECK(CheckParseUtf8Char("\xfd\x80\x80\x80\x80\x80", 1 << 30), "6 bytes");

   CHECK(CheckParseUtf8Char("\xc0", 0xfffd), "bad 2 bytes");
   CHECK(CheckParseUtf8Char("\xe0\x80", 0xfffd), "bad 3 bytes");
   CHECK(CheckParseUtf8Char("\xf0\x80\x80", 0xfffd), "bad 4 bytes");
   CHECK(CheckParseUtf8Char("\xf8\x80\x80\x80", 0xfffd), "bad 5 bytes");
   CHECK(CheckParseUtf8Char("\xfc\x80\x80\x80\x80", 0xfffd), "bad 6 bytes");

   CHECK(CheckParseUtf8Char("\xdf\xb7", 0x7f7), "2 bytes");
   CHECK(CheckParseUtf8Char("\xe2\x98\x83", 0x2603), "3 bytes");
   CHECK(CheckParseUtf8Char("\xf0\x90\x83\xa1", 0x100e1), "4 bytes");
   CHECK(CheckParseUtf8Char("\xfa\x80\x80\x80\x80", 0x2000000), "5 bytes");
   CHECK(CheckParseUtf8Char("\xfd\x80\x80\x80\x80\x80", 0x40000000), "6 bytes");
}

// Try skipping to the next character
static void TestNextChar()
{
   string text =
      "a"
      "\xd0\x80"
      "\xe8\x80\x80"
      "\xf4\x80\x80\x80"
      "\xfa\x80\x80\x80\x80"
      "\xfd\x80\x80\x80\x80\x80"
      "z";
   string::iterator i = text.begin();

   #define CHECK_ITERATOR(expected_char, label) \
      NextChar(&text, &i);                      \
      CHECK(i != text.end(), label);            \
      CHECK(*i == expected_char, label)
   CHECK_ITERATOR('\xd0', "1 byte");
   CHECK_ITERATOR('\xe8', "2 bytes");
   CHECK_ITERATOR('\xf4', "3 bytes");
   CHECK_ITERATOR('\xfa', "4 bytes");
   CHECK_ITERATOR('\xfd', "5 bytes");
   CHECK_ITERATOR('z', "6 bytes");
   #undef CHECK_ITERATOR

   NextChar(&text, &i);
   CHECK(i == text.end(), "1 byte");
}

}  // namespace

int main()
{
   TestParseUtf8();
   TestNextChar();
   return 0;
}