fmt,cxxrtl: add UNICHAR format type.

This format type is used to print an Unicode character (code point) as
its UTF-8 serialization. To this end, two UTF-8 decoders (one for fmt,
one for cxxrtl) are added for rendering. When converted to a Verilog
format specifier, `UNICHAR` degrades to `%c` with the low 7 bits of
the code point, which has equivalent behavior for inputs not exceeding
ASCII. (SystemVerilog leaves source and display encodings completely
undefined.)

backends/cxxrtl/runtime/cxxrtl/cxxrtl.h

@@ -1013,13 +1013,14 @@ struct fmt_part {
 		LITERAL   = 0,
 		INTEGER   = 1,
 		STRING    = 2,
-		VLOG_TIME = 3,
+		UNICHAR   = 3,
+		VLOG_TIME = 4,
 	} type;
 
 	// LITERAL type
 	std::string str;
 
-	// INTEGER/STRING types
+	// INTEGER/STRING/UNICHAR types
 	// + value<Bits> val;
 
 	// INTEGER/STRING/VLOG_TIME types
@@ -1073,6 +1074,25 @@ struct fmt_part {
 				break;
 			}
 
+			case UNICHAR: {
+				uint32_t codepoint = val.template get<uint32_t>();
+				if (codepoint >= 0x10000)
+					buf += (char)(0xf0 |  (codepoint >> 18));
+				else if (codepoint >= 0x800)
+					buf += (char)(0xe0 |  (codepoint >> 12));
+				else if (codepoint >= 0x80)
+					buf += (char)(0xc0 |  (codepoint >>  6));
+				else
+					buf += (char)codepoint;
+				if (codepoint >= 0x10000)
+					buf += (char)(0x80 | ((codepoint >> 12) & 0x3f));
+				if (codepoint >= 0x800)
+					buf += (char)(0x80 | ((codepoint >>  6) & 0x3f));
+				if (codepoint >= 0x80)
+					buf += (char)(0x80 | ((codepoint >>  0) & 0x3f));
+				break;
+			}
+
 			case INTEGER: {
 				size_t width = Bits;
 				if (base != 10) {

kernel/fmt.cc

@@ -42,9 +42,9 @@ void Fmt::parse_rtlil(const RTLIL::Cell *cell) {
 		} else if (fmt.substr(i, 2) == "{{") {
 			part.str += '{';
 			++i;
-		} else if (fmt[i] == '}')
+		} else if (fmt[i] == '}') {
 			log_assert(false && "Unexpected '}' in format string");
-		else if (fmt[i] == '{') {
+		} else if (fmt[i] == '{') {
 			if (!part.str.empty()) {
 				part.type = FmtPart::LITERAL;
 				parts.push_back(part);
@@ -74,6 +74,12 @@ void Fmt::parse_rtlil(const RTLIL::Cell *cell) {
 			part.sig = args.extract(0, arg_size);
 			args.remove(0, arg_size);
 
+			if (fmt[i] == 'U') {
+				part.type = FmtPart::UNICHAR;
+				++i;
+				goto success;
+			}
+
 			if (fmt[i] == '>')
 				part.justify = FmtPart::RIGHT;
 			else if (fmt[i] == '<')
@@ -156,6 +162,7 @@ void Fmt::parse_rtlil(const RTLIL::Cell *cell) {
 					log_assert(false && "Unexpected end in format substitution");
 			}
 
+		success:
 			if (fmt[i] != '}')
 				log_assert(false && "Expected '}' after format substitution");
 
@@ -188,6 +195,11 @@ void Fmt::emit_rtlil(RTLIL::Cell *cell) const {
 				}
 				break;
 
+			case FmtPart::UNICHAR:
+				log_assert(part.sig.size() <= 32);
+				fmt += "{U}";
+				break;
+
 			case FmtPart::VLOG_TIME:
 				log_assert(part.sig.size() == 0);
 				YS_FALLTHROUGH
@@ -568,6 +580,16 @@ std::vector<VerilogFmtArg> Fmt::emit_verilog() const
 				break;
 			}
 
+			case FmtPart::UNICHAR: {
+				VerilogFmtArg arg;
+				arg.type = VerilogFmtArg::INTEGER;
+				arg.sig = part.sig.extract(0, 7); // only ASCII
+				args.push_back(arg);
+
+				fmt.str += "%c";
+				break;
+			}
+
 			case FmtPart::VLOG_TIME: {
 				VerilogFmtArg arg;
 				arg.type = VerilogFmtArg::TIME;
@@ -630,6 +652,7 @@ void Fmt::emit_cxxrtl(std::ostream &os, std::string indent, std::function<void(c
 			case FmtPart::LITERAL:   os << "LITERAL";   break;
 			case FmtPart::INTEGER:   os << "INTEGER";   break;
 			case FmtPart::STRING:    os << "STRING";    break;
+			case FmtPart::UNICHAR:   os << "UNICHAR";   break;
 			case FmtPart::VLOG_TIME: os << "VLOG_TIME"; break;
 		}
 		os << ", ";
@@ -671,6 +694,26 @@ std::string Fmt::render() const
 				str += part.str;
 				break;
 
+			case FmtPart::UNICHAR: {
+				RTLIL::Const value = part.sig.as_const();
+				uint32_t codepoint = value.as_int();
+				if (codepoint >= 0x10000)
+					str += (char)(0xf0 |  (codepoint >> 18));
+				else if (codepoint >= 0x800)
+					str += (char)(0xe0 |  (codepoint >> 12));
+				else if (codepoint >= 0x80)
+					str += (char)(0xc0 |  (codepoint >>  6));
+				else
+					str += (char)codepoint;
+				if (codepoint >= 0x10000)
+					str += (char)(0x80 | ((codepoint >> 12) & 0x3f));
+				if (codepoint >= 0x800)
+					str += (char)(0x80 | ((codepoint >>  6) & 0x3f));
+				if (codepoint >= 0x80)
+					str += (char)(0x80 | ((codepoint >>  0) & 0x3f));
+				break;
+			}
+
 			case FmtPart::INTEGER:
 			case FmtPart::STRING:
 			case FmtPart::VLOG_TIME: {

kernel/fmt.h

@@ -56,13 +56,14 @@ struct FmtPart {
 		LITERAL  	= 0,
 		INTEGER 	= 1,
 		STRING    = 2,
-		VLOG_TIME = 3,
+		UNICHAR   = 3,
+		VLOG_TIME = 4,
 	} type;
 
 	// LITERAL type
 	std::string str;
 
-	// INTEGER/STRING types
+	// INTEGER/STRING/UNICHAR types
 	RTLIL::SigSpec sig;
 
 	// INTEGER/STRING/VLOG_TIME types