Adding error case tests for UTF-16 to UTF-8 conversions. Little bit of error handling cleanup as well.

Author: stgates

Release/src/utilities/asyncrt_utils.cpp

@@ -294,12 +294,12 @@ utf16string __cdecl conversions::utf8_to_utf16(const std::string &s)
             }
             else
             {
-                throw std::invalid_argument("UTF-8 string has invalid Unicode code point");
+                throw std::range_error("UTF-8 string has invalid Unicode code point");
             }
             srcRemainingSize -= numContBytes;
             if (srcRemainingSize <= 0)
             {
-                throw std::invalid_argument("UTF-8 string is missing bytes in character");
+                throw std::range_error("UTF-8 string is missing bytes in character");
             }
 
             for (unsigned char i = 0; i < numContBytes; ++i)
@@ -341,18 +341,18 @@ std::string __cdecl conversions::utf16_to_utf8(const utf16string &w)
      return conversion.to_bytes(w);
  #else
     std::string dest;
-    dest.reserve(w.size()); // TODO size
+    dest.reserve(w.size());
     const utf16string::value_type *src = w.c_str();
     auto srcRemainingSize = w.size();
     while (srcRemainingSize > 0)
     {
+        // Check for high surrogate.
         if (*src >= 0xD800 && *src <= 0xDBFF)
         {
             if (--srcRemainingSize == 0)
             {
-                // TODO error
+                throw std::range_error("UTF-16 string is missing low surrogate");
             }
-            // Found a high surrogate.
 
             // To get from surrogate pair to Unicode code point:
             // - subract 0xD800 from high surrogate, this forms top ten bits
@@ -363,6 +363,10 @@ std::string __cdecl conversions::utf16_to_utf8(const utf16string &w)
             codePoint <<= 10;
             codePoint += *++src - 0xDC00;
             codePoint += 0x10000;
+            if (*src < 0xDC00 || *src > 0xDFFF)
+            {
+                throw std::range_error("UTF-16 string has invalid low surrogate");
+            }
 
             // 4 bytes need using 21 bits
             dest.push_back(char(codePoint >> 18) | 0xF0);               // leading 3 bits

Release/tests/functional/utils/strings.cpp

@@ -191,32 +191,48 @@ TEST(utf8_to_utf16)
 #endif
 }
 
+TEST(utf16_to_utf8_errors)
+{
+    VERIFY_ARE_EQUAL("ABC987", utility::conversions::utf16_to_utf8(UTF16("ABC987")));
+    utf16string input;
+    
+    // high surrogate with missing low surrogate.
+    input.push_back(0xD800);
+    input.push_back(0x0);
+    VERIFY_THROWS(utility::conversions::utf16_to_utf8(input), std::range_error);
+
+    // high surrogate with no more characters
+    input.clear();
+    input.push_back(0xD800);
+    VERIFY_THROWS(utility::conversions::utf16_to_utf8(input), std::range_error);
+}
+
 TEST(utf8_to_utf16_errors)
 {
     // missing second continuation byte
     std::string input;
     input.push_back(207u); // 11001111
-    VERIFY_THROWS(utility::conversions::utf8_to_utf16(input), std::invalid_argument);
+    VERIFY_THROWS(utility::conversions::utf8_to_utf16(input), std::range_error);
 
     // missing third continuation byte
     input.clear();
     input.push_back(230u); // 11100110
     input.push_back(141u); // 10001101
-    VERIFY_THROWS(utility::conversions::utf8_to_utf16(input), std::invalid_argument);
+    VERIFY_THROWS(utility::conversions::utf8_to_utf16(input), std::range_error);
 
     // missing fourth continuation byte
     input.clear();
     input.push_back(240u); // 11110000
     input.push_back(173u); // 10101101
     input.push_back(157u); // 10011101
-    VERIFY_THROWS(utility::conversions::utf8_to_utf16(input), std::invalid_argument);
+    VERIFY_THROWS(utility::conversions::utf8_to_utf16(input), std::range_error);
 }
 
 TEST(latin1_to_utf16)
 {
     // TODO: find some string that actually uses something unique to the Latin1 code page.
     std::string str_latin1("This is a test");
-    utf16string str_utf16 = utility::conversions::usascii_to_utf16(str_latin1);
+    utf16string str_utf16 = utility::conversions::latin1_to_utf16(str_latin1);
 
     for (size_t i = 0; i < str_latin1.size(); ++i)
     {