So JSON is unicode, nothing to do with UTF8

SmingHub · Aug 31, 2024 · e8c2dee · e8c2dee
1 parent 1905dda
commit e8c2dee
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 71 deletions.
diff --git a/Sming/Core/Data/Format/Formatter.cpp b/Sming/Core/Data/Format/Formatter.cpp
@@ -57,11 +57,15 @@ unsigned escapeControls(String& value, Options options)
 	for(auto& c : value) {
 		if(escapeChar(c, options)) {
 			extra += 1; // "\"
+		} else if(options[Option::unicode]) {
+			if(uint8_t(c) < 0x20 || (c & 0x80)) {
+				extra += 5; // "\uNNNN"
+			}
 		} else if(uint8_t(c) < 0x20) {
-			extra += options[Option::unicode] ? 5 : 3; // "\uNNNN" or "\xnn"
-		} else if(options[Option::utf8] && (c & 0x80)) {
+			extra += 3; // "\xnn"
+		} else if((c & 0x80) && options[Option::utf8]) {
 			// Characters such as £ (0xa3) are escaped to 0xc2 0xa3 in UTF-8
-			extra += 1; // '\xc2' prefix
+			extra += 1; // 0xc2
 		}
 	}
 	if(extra == 0) {
@@ -80,24 +84,25 @@ unsigned escapeControls(String& value, Options options)
 		auto esc = escapeChar(c, options);
 		if(esc) {
 			*out++ = '\\';
-			*out++ = esc;
-		} else if(c < 0x20) {
-			*out++ = '\\';
-			if(options[Option::unicode]) {
+			c = esc;
+		} else if(options[Option::unicode]) {
+			if(uint8_t(c) < 0x20 || (c & 0x80)) {
+				*out++ = '\\';
 				*out++ = 'u';
 				*out++ = '0';
 				*out++ = '0';
-			} else {
-				*out++ = 'x';
+				*out++ = hexchar(uint8_t(c) >> 4);
+				c = hexchar(uint8_t(c) & 0x0f);
 			}
+		} else if(uint8_t(c) < 0x20) {
+			*out++ = '\\';
+			*out++ = 'x';
 			*out++ = hexchar(uint8_t(c) >> 4);
-			*out++ = hexchar(uint8_t(c) & 0x0f);
-		} else if(options[Option::utf8] && (c & 0x80)) {
-			*out++ = '\xc2';
-			*out++ = c;
-		} else {
-			*out++ = c;
+			c = hexchar(uint8_t(c) & 0x0f);
+		} else if((c & 0x80) && options[Option::utf8]) {
+			*out++ = 0xc2;
 		}
+		*out++ = c;
 	}
 	return extra;
 }

diff --git a/Sming/Core/Data/Format/Json.cpp b/Sming/Core/Data/Format/Json.cpp
@@ -17,59 +17,6 @@ namespace Format
 {
 Json json;
 
-namespace
-{
-bool IsValidUtf8(const char* str, unsigned length)
-{
-	if(str == nullptr) {
-		return true;
-	}
-
-	unsigned i = 0;
-	while(i < length) {
-		char c = str[i++];
-		if((c & 0x80) == 0) {
-			continue;
-		}
-
-		if(i >= length) {
-			return false; // incomplete multibyte char
-		}
-
-		if(c & 0x20) {
-			c = str[i++];
-			if((c & 0xC0) != 0x80) {
-				return false; // malformed trail byte or out of range char
-			}
-			if(i >= length) {
-				return false; // incomplete multibyte char
-			}
-		}
-
-		c = str[i++];
-		if((c & 0xC0) != 0x80) {
-			return false; // malformed trail byte
-		}
-	}
-
-	return true;
-}
-
-void escapeText(String& value)
-{
-	escapeControls(value, Option::unicode | Option::utf8 | Option::doublequote | Option::backslash);
-	if(!IsValidUtf8(value.c_str(), value.length())) {
-		debug_w("Invalid UTF8: %s", value.c_str());
-		for(unsigned i = 0; i < value.length(); ++i) {
-			char& c = value[i];
-			if(c < 0x20 || uint8_t(c) > 127)
-				c = '_';
-		}
-	}
-}
-
-} // namespace
-
 /*
  * Check for invalid characters and replace them - can break browser
  * operation otherwise.
@@ -79,12 +26,12 @@ void escapeText(String& value)
  */
 void Json::escape(String& value) const
 {
-	escapeText(value);
+	escapeControls(value, Option::unicode | Option::doublequote | Option::backslash);
 }
 
 void Json::quote(String& value) const
 {
-	escapeText(value);
+	escape(value);
 	auto len = value.length();
 	if(value.setLength(len + 2)) {
 		auto s = value.begin();

diff --git a/tests/HostTests/modules/Formatter.cpp b/tests/HostTests/modules/Formatter.cpp
@@ -17,7 +17,7 @@ class FormatterTest : public TestGroup
 		TEST_CASE("JSON")
 		{
 			DEFINE_FSTR_LOCAL(text1b, "A JSON\\ntest string\\twith escapes\\u0012\\u0000\\n"
-									  "Worth \\\"maybe\\\" \xc2\xa3 0.53. Yen \xc2\xa5 5bn.")
+									  "Worth \\\"maybe\\\" \\u00a3 0.53. Yen \\u00a5 5bn.")
 
 			Serial << text1 << endl;
 			String s(text1);