Add new view extension cstring function tests

Author: Aidan Lee

src/cpp/encoding/Utf16.cpp

@@ -19,6 +19,42 @@ namespace
 	{
 		return codepoint >= 0xd800 && codepoint < 0xdc00;
 	}
+
+	bool isAsciiBuffer(View<uint8_t> buffer)
+	{
+		while (buffer.isEmpty() == false)
+		{
+			auto p = cpp::encoding::Utf16::codepoint(buffer);
+
+			if (p > 127)
+			{
+				return false;
+			}
+
+			buffer = buffer.slice(cpp::encoding::Utf16::getByteCount(p));
+		}
+
+		return true;
+	}
+
+	String toAsciiString(View<uint8_t> buffer)
+	{
+		auto bytes  = buffer.length / sizeof(char16_t);
+		auto chars  = View<char>(hx::InternalNew(bytes + 1, false), bytes * sizeof(char));
+		auto output = chars.reinterpret<uint8_t>();
+
+		while (buffer.isEmpty() == false)
+		{
+			auto p = cpp::encoding::Utf16::codepoint(buffer);
+
+			output[0] = static_cast<uint8_t>(p);
+
+			buffer = buffer.slice(cpp::encoding::Utf16::getByteCount(p));
+			output = output.slice(1);
+		}
+		
+		return String(chars.ptr.ptr, chars.length);
+	}
 }
 
 bool cpp::encoding::Utf16::isEncoded(const String& string)
@@ -174,6 +210,11 @@ String cpp::encoding::Utf16::decode(cpp::marshal::View<uint8_t> buffer)
 		return String::emptyString;
 	}
 
+	if (isAsciiBuffer(buffer))
+	{
+		return toAsciiString(buffer);
+	}
+
     auto chars = int64_t{ 0 };
 	auto i     = int64_t{ 0 };
     while (i < buffer.length)

src/cpp/encoding/Utf8.cpp

@@ -5,14 +5,18 @@ using namespace cpp::marshal;
 
 namespace
 {
-    bool isAsciiBuffer(View<uint8_t>& buffer)
+    bool isAsciiBuffer(View<uint8_t> buffer)
     {
-        for (auto i = int64_t{ 0 }; i < buffer.length; i++)
+        while (buffer.isEmpty() == false)
         {
-            if (buffer.ptr[i] > 127)
+            auto p = cpp::encoding::Utf8::codepoint(buffer);
+
+            if (p > 127)
             {
                 return false;
             }
+
+            buffer = buffer.slice(cpp::encoding::Utf8::getByteCount(p));
         }
 
         return true;

test/native/cpp/encoding/Ascii.hx

@@ -141,23 +141,17 @@ class TestUtf16 extends Test {
 	}
 
 	public function test_decode_codepoint() {
-		var codepoint : cpp.Char32 = 0;
-
 		var bytes = Bytes.ofHex('6100');
-		Assert.equals(2i64, Utf16.decode(bytes.asView(), codepoint));
-		Assert.equals('a'.code, cast codepoint);
+		Assert.equals('a'.code, Utf16.codepoint(bytes.asView()));
 
 		var bytes = Bytes.ofHex('8501');
-		Assert.equals(2i64, Utf16.decode(bytes.asView(), codepoint));
-		Assert.equals('ƅ'.code, cast codepoint);
+		Assert.equals('ƅ'.code, Utf16.codepoint(bytes.asView()));
 
 		var bytes = Bytes.ofHex('D030');
-		Assert.equals(2i64, Utf16.decode(bytes.asView(), codepoint));
-		Assert.equals('バ'.code, cast codepoint);
+		Assert.equals('バ'.code, Utf16.codepoint(bytes.asView()));
 
 		var bytes = Bytes.ofHex('34D833DD');
-		Assert.equals(4i64, Utf16.decode(bytes.asView(), codepoint));
-		Assert.equals('𝄳'.code, cast codepoint);
+		Assert.equals('𝄳'.code, Utf16.codepoint(bytes.asView()));
 	}
 
 	public function test_decode_string() {

test/native/cpp/encoding/Utf16.hx

@@ -118,23 +118,17 @@ class TestUtf8 extends Test {
 	}
 
 	public function test_decode_codepoint() {
-		var codepoint : cpp.Char32 = 0;
-
 		var bytes = Bytes.ofHex('61');
-		Assert.equals(1i64, Utf8.decode(bytes.asView(), codepoint));
-		Assert.equals('a'.code, cast codepoint);
+		Assert.equals('a'.code, Utf8.codepoint(bytes.asView()));
 
 		var bytes = Bytes.ofHex('c685');
-		Assert.equals(2i64, Utf8.decode(bytes.asView(), codepoint));
-		Assert.equals('ƅ'.code, cast codepoint);
+		Assert.equals('ƅ'.code, Utf8.codepoint(bytes.asView()));
 
 		var bytes = Bytes.ofHex('e38390');
-		Assert.equals(3i64, Utf8.decode(bytes.asView(), codepoint));
-		Assert.equals('バ'.code, cast codepoint);
+		Assert.equals('バ'.code, Utf8.codepoint(bytes.asView()));
 
 		var bytes = Bytes.ofHex('f09d84b3');
-		Assert.equals(4i64, Utf8.decode(bytes.asView(), codepoint));
-		Assert.equals('𝄳'.code, cast codepoint);
+		Assert.equals('𝄳'.code, Utf8.codepoint(bytes.asView()));
 	}
 
 	public function test_decode_string() {

test/native/cpp/encoding/Utf8.hx

@@ -162,4 +162,54 @@ class TestViewExtensions extends Test {
 			}
 		}
 	}
+
+	function test_szToString_char_no_null() {
+		final vec = new Vector<cpp.Char>(4);
+        vec[0] = 't'.code;
+        vec[1] = 'e'.code;
+        vec[2] = 's'.code;
+        vec[3] = 't'.code;
+
+		Assert.equals("test", vec.asView().szToString());
+	}
+
+	function test_szToString_char() {
+		final vec = new Vector<cpp.Char>(9);
+        vec[0] = 't'.code;
+        vec[1] = 'e'.code;
+        vec[2] = 's'.code;
+        vec[3] = 't'.code;
+        vec[4] = 0;
+        vec[5] = 't'.code;
+        vec[6] = 'e'.code;
+        vec[7] = 's'.code;
+        vec[8] = 't'.code;
+
+		Assert.equals("test", vec.asView().szToString());
+	}
+
+	function test_szToString_char16_no_null() {
+		final vec = new Vector<cpp.Char16>(4);
+        vec[0] = 't'.code;
+        vec[1] = 'e'.code;
+        vec[2] = 's'.code;
+        vec[3] = 't'.code;
+
+		Assert.equals("test", vec.asView().szToString());
+	}
+
+	function test_szToString16_char() {
+		final vec = new Vector<cpp.Char16>(9);
+        vec[0] = 't'.code;
+        vec[1] = 'e'.code;
+        vec[2] = 's'.code;
+        vec[3] = 't'.code;
+        vec[4] = 0;
+        vec[5] = 't'.code;
+        vec[6] = 'e'.code;
+        vec[7] = 's'.code;
+        vec[8] = 't'.code;
+
+		Assert.equals("test", vec.asView().szToString());
+	}
 }