[string] string.ToLower, string.ToUpper in C++ using ICU.

#143
fusionlanguage · Feb 19, 2024 · 82c353d · 82c353d
1 parent 0c75701
commit 82c353d
Show file tree

Hide file tree

Showing 8 changed files with 58 additions and 77 deletions.
diff --git a/GenCpp.fu b/GenCpp.fu
@@ -568,6 +568,15 @@ public class GenCpp : GenCCpp
 			WriteArgsInParentheses(method, args);
 	}
 
+	void WriteStringToLowerUpper!(FuExpr obj, string name)
+	{
+		Include("string");
+		Include("unicode/unistr.h");
+		Write("[](icu::StringPiece s) { std::string result; return icu::UnicodeString::fromUTF8(s).to");
+		Write(name);
+		WriteCall("er().toUTF8String(result); }", obj);
+	}
+
 	void WriteAllAnyContains!(string function, FuExpr obj, List<FuExpr#> args)
 	{
 		Include("algorithm");
@@ -809,24 +818,10 @@ public class GenCpp : GenCCpp
 			WriteStringMethod(obj, "substr", method, args);
 			break;
 		case FuId.StringToLower:
-			Include("algorithm");
-			Include("cctype");
-			Write("[&] { std::string data = std::string{");
-			obj.Accept(this, FuPriority.Argument);
-			Write("}; ");
-			Write("std::transform(data.begin(), data.end(), data.begin(), ");
-			Write("[](unsigned char c) { return std::tolower(c); }); ");
-			Write("return data; }()");
+			WriteStringToLowerUpper(obj, "Low");
 			break;
 		case FuId.StringToUpper:
-			Include("algorithm");
-			Include("cctype");
-			Write("[&] { std::string data = std::string{");
-			obj.Accept(this, FuPriority.Argument);
-			Write("}; ");
-			Write("std::transform(data.begin(), data.end(), data.begin(), ");
-			Write("[](unsigned char c) { return std::toupper(c); }); ");
-			Write("return data; }()");
+			WriteStringToLowerUpper(obj, "Upp");
 			break;
 		case FuId.ArrayBinarySearchAll:
 		case FuId.ArrayBinarySearchPart:

diff --git a/Makefile b/Makefile
@@ -181,7 +181,7 @@ test/bin/%/c.exe: test/bin/%/Test.c
 	$(DO)$(CC) -o $@ $(TEST_CFLAGS) -Wno-unused-function -I $(<D) $^ `pkg-config --cflags --libs glib-2.0` -lm || grep '//FAIL:.*\<c\>' test/$*.fu
 
 test/bin/%/cpp.exe: test/bin/%/Test.cpp
-	$(DO)$(CXX) -o $@ $(TEST_CXXFLAGS) -I $(<D) $^ || grep '//FAIL:.*\<cpp\>' test/$*.fu
+	$(DO)$(CXX) -o $@ $(TEST_CXXFLAGS) -I $(<D) $^ $(if $(findstring $*, StringToLower StringToUpper), -licuuc) || grep '//FAIL:.*\<cpp\>' test/$*.fu
 
 test/bin/%/cs.dll: test/bin/%/Test.cs
 	$(DO)$(CSC) -out:$@ $^ || grep '//FAIL:.*\<cs\>' test/$*.fu

diff --git a/libfut.cpp b/libfut.cpp
@@ -13532,6 +13532,15 @@ void GenCpp::writeStringMethod(const FuExpr * obj, std::string_view name, const
 		writeArgsInParentheses(method, args);
 }
 
+void GenCpp::writeStringToLowerUpper(const FuExpr * obj, std::string_view name)
+{
+	include("string");
+	include("unicode/unistr.h");
+	write("[](icu::StringPiece s) { std::string result; return icu::UnicodeString::fromUTF8(s).to");
+	write(name);
+	writeCall("er().toUTF8String(result); }", obj);
+}
+
 void GenCpp::writeAllAnyContains(std::string_view function, const FuExpr * obj, const std::vector<std::shared_ptr<FuExpr>> * args)
 {
 	include("algorithm");
@@ -13770,24 +13779,10 @@ void GenCpp::writeCallExpr(const FuExpr * obj, const FuMethod * method, const st
 		writeStringMethod(obj, "substr", method, args);
 		break;
 	case FuId::stringToLower:
-		include("algorithm");
-		include("cctype");
-		write("[&] { std::string data = std::string{");
-		obj->accept(this, FuPriority::argument);
-		write("}; ");
-		write("std::transform(data.begin(), data.end(), data.begin(), ");
-		write("[](unsigned char c) { return std::tolower(c); }); ");
-		write("return data; }()");
+		writeStringToLowerUpper(obj, "Low");
 		break;
 	case FuId::stringToUpper:
-		include("algorithm");
-		include("cctype");
-		write("[&] { std::string data = std::string{");
-		obj->accept(this, FuPriority::argument);
-		write("}; ");
-		write("std::transform(data.begin(), data.end(), data.begin(), ");
-		write("[](unsigned char c) { return std::toupper(c); }); ");
-		write("return data; }()");
+		writeStringToLowerUpper(obj, "Upp");
 		break;
 	case FuId::arrayBinarySearchAll:
 	case FuId::arrayBinarySearchPart:

diff --git a/libfut.cs b/libfut.cs
@@ -13882,6 +13882,15 @@ void WriteStringMethod(FuExpr obj, string name, FuMethod method, List<FuExpr> ar
 				WriteArgsInParentheses(method, args);
 		}
 
+		void WriteStringToLowerUpper(FuExpr obj, string name)
+		{
+			Include("string");
+			Include("unicode/unistr.h");
+			Write("[](icu::StringPiece s) { std::string result; return icu::UnicodeString::fromUTF8(s).to");
+			Write(name);
+			WriteCall("er().toUTF8String(result); }", obj);
+		}
+
 		void WriteAllAnyContains(string function, FuExpr obj, List<FuExpr> args)
 		{
 			Include("algorithm");
@@ -14119,24 +14128,10 @@ protected override void WriteCallExpr(FuExpr obj, FuMethod method, List<FuExpr>
 				WriteStringMethod(obj, "substr", method, args);
 				break;
 			case FuId.StringToLower:
-				Include("algorithm");
-				Include("cctype");
-				Write("[&] { std::string data = std::string{");
-				obj.Accept(this, FuPriority.Argument);
-				Write("}; ");
-				Write("std::transform(data.begin(), data.end(), data.begin(), ");
-				Write("[](unsigned char c) { return std::tolower(c); }); ");
-				Write("return data; }()");
+				WriteStringToLowerUpper(obj, "Low");
 				break;
 			case FuId.StringToUpper:
-				Include("algorithm");
-				Include("cctype");
-				Write("[&] { std::string data = std::string{");
-				obj.Accept(this, FuPriority.Argument);
-				Write("}; ");
-				Write("std::transform(data.begin(), data.end(), data.begin(), ");
-				Write("[](unsigned char c) { return std::toupper(c); }); ");
-				Write("return data; }()");
+				WriteStringToLowerUpper(obj, "Upp");
 				break;
 			case FuId.ArrayBinarySearchAll:
 			case FuId.ArrayBinarySearchPart:

diff --git a/libfut.hpp b/libfut.hpp
@@ -2320,6 +2320,7 @@ class GenCpp : public GenCCpp
 	void writePtrRange(const FuExpr * obj, const FuExpr * index, const FuExpr * count);
 	void writeNotRawStringLiteral(const FuExpr * obj, FuPriority priority);
 	void writeStringMethod(const FuExpr * obj, std::string_view name, const FuMethod * method, const std::vector<std::shared_ptr<FuExpr>> * args);
+	void writeStringToLowerUpper(const FuExpr * obj, std::string_view name);
 	void writeAllAnyContains(std::string_view function, const FuExpr * obj, const std::vector<std::shared_ptr<FuExpr>> * args);
 	void writeCollectionMethod(const FuExpr * obj, std::string_view name, const std::vector<std::shared_ptr<FuExpr>> * args);
 	void writeCString(const FuExpr * expr);

diff --git a/libfut.js b/libfut.js
@@ -14312,6 +14312,15 @@ export class GenCpp extends GenCCpp
 			this.writeArgsInParentheses(method, args);
 	}
 
+	#writeStringToLowerUpper(obj, name)
+	{
+		this.include("string");
+		this.include("unicode/unistr.h");
+		this.write("[](icu::StringPiece s) { std::string result; return icu::UnicodeString::fromUTF8(s).to");
+		this.write(name);
+		this.writeCall("er().toUTF8String(result); }", obj);
+	}
+
 	#writeAllAnyContains(function_, obj, args)
 	{
 		this.include("algorithm");
@@ -14551,24 +14560,10 @@ export class GenCpp extends GenCCpp
 			this.#writeStringMethod(obj, "substr", method, args);
 			break;
 		case FuId.STRING_TO_LOWER:
-			this.include("algorithm");
-			this.include("cctype");
-			this.write("[&] { std::string data = std::string{");
-			obj.accept(this, FuPriority.ARGUMENT);
-			this.write("}; ");
-			this.write("std::transform(data.begin(), data.end(), data.begin(), ");
-			this.write("[](unsigned char c) { return std::tolower(c); }); ");
-			this.write("return data; }()");
+			this.#writeStringToLowerUpper(obj, "Low");
 			break;
 		case FuId.STRING_TO_UPPER:
-			this.include("algorithm");
-			this.include("cctype");
-			this.write("[&] { std::string data = std::string{");
-			obj.accept(this, FuPriority.ARGUMENT);
-			this.write("}; ");
-			this.write("std::transform(data.begin(), data.end(), data.begin(), ");
-			this.write("[](unsigned char c) { return std::toupper(c); }); ");
-			this.write("return data; }()");
+			this.#writeStringToLowerUpper(obj, "Upp");
 			break;
 		case FuId.ARRAY_BINARY_SEARCH_ALL:
 		case FuId.ARRAY_BINARY_SEARCH_PART:

diff --git a/test/StringToLower.fu b/test/StringToLower.fu
@@ -1,4 +1,4 @@
-// https://pastebin.com/fuw4Uizk
+// https://pastebin.com/fuw4Uizk
 // ĐÂĂÊÔƠƯẤẮẾỐỚỨẦẰỀỒỜỪẬẶỆỘỢỰШ ΟΔΥΣΣΕΥΣ
 // đâăêôơưấắếốớứầằềồờừậặệộợựш ὀδυσσεύς
 
@@ -8,11 +8,11 @@ public class Test
 	{
 		string() orig1 = "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш"; //FAIL: cl
 		string orig2 = "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш";
-		string() new1 = orig1.ToLower();
+		string() new1 = orig1.ToLower(); //FAIL: cpp requires ICU
 		string() new2 = orig2.ToLower();
-		return orig1 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш" &&
-			orig2 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш" &&
-			new1 == "mixed case đâăêôơưấắếốớứầằềồờừậặệộợựш" &&
-			new2 == "mixed case đâăêôơưấắếốớứầằềồờừậặệộợựш";
+		return orig1 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш"
+			&& orig2 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш"
+			&& new1 == "mixed case đâăêôơưấắếốớứầằềồờừậặệộợựш"
+			&& new2 == "mixed case đâăêôơưấắếốớứầằềồờừậặệộợựш";
 	}
 }
diff --git a/test/StringToUpper.fu b/test/StringToUpper.fu
@@ -1,4 +1,4 @@
-// https://pastebin.com/fuw4Uizk
+// https://pastebin.com/fuw4Uizk
 // ĐÂĂÊÔƠƯẤẮẾỐỚỨẦẰỀỒỜỪẬẶỆỘỢỰШ
 // đâăêôơưấắếốớứầằềồờừậặệộợựш
 
@@ -8,11 +8,11 @@ public class Test
 	{
 		string() orig1 = "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш"; //FAIL: cl
 		string orig2 = "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш";
-		string() new1 = orig1.ToUpper();
+		string() new1 = orig1.ToUpper(); //FAIL: cpp requires ICU
 		string() new2 = orig2.ToUpper();
-		return orig1 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш" &&
-			orig2 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш" &&
-			new1 == "MIXED CASE ĐÂĂÊÔƠƯẤẮẾỐỚỨẦẰỀỒỜỪẬẶỆỘỢỰШ" &&
-			new2 == "MIXED CASE ĐÂĂÊÔƠƯẤẮẾỐỚỨẦẰỀỒỜỪẬẶỆỘỢỰШ";
+		return orig1 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш"
+			&& orig2 == "MiXeD cAsE ĐÂĂÊÔơưấắếỐỚỨẦẰềồờừậẶỆỘỢỰш"
+			&& new1 == "MIXED CASE ĐÂĂÊÔƠƯẤẮẾỐỚỨẦẰỀỒỜỪẬẶỆỘỢỰШ"
+			&& new2 == "MIXED CASE ĐÂĂÊÔƠƯẤẮẾỐỚỨẦẰỀỒỜỪẬẶỆỘỢỰШ";
 	}
 }