Feature [feature-requests:#1408] Use positive IsDBCSTrailByteNoExcept function

instead of negated IsDBCSTrailByteInvalid.

3 files changed, 46 insertions, 2 deletions

diff --git a/src/Document.cxx b/src/Document.cxx
index 876e95b4f..739b6266a 100644
--- a/src/Document.cxx
+++ b/src/Document.cxx
@@ -894,7 +894,7 @@ Document::CharacterExtracted Document::CharacterAfter(Sci::Position position) co
 	} else {
 		if (IsDBCSLeadByteNoExcept(leadByte)) {
 			const unsigned char trailByte = cb.UCharAt(position + 1);
-			if (!IsDBCSTrailByteInvalid(trailByte)) {
+			if (IsDBCSTrailByteNoExcept(trailByte)) {
 				return CharacterExtracted::DBCS(leadByte, trailByte);
 			}
 		}
@@ -1010,7 +1010,7 @@ int SCI_METHOD Document::GetCharacterAndWidth(Sci_Position position, Sci_Positio
 		} else {
 			if (IsDBCSLeadByteNoExcept(leadByte)) {
 				const unsigned char trailByte = cb.UCharAt(position + 1);
-				if (!IsDBCSTrailByteInvalid(trailByte)) {
+				if (IsDBCSTrailByteNoExcept(trailByte)) {
 					bytesInCharacter = 2;
 					character = (leadByte << 8) | trailByte;
 				} else {
@@ -1067,6 +1067,37 @@ bool Document::IsDBCSLeadByteNoExcept(char ch) const noexcept {
 	return false;
 }
 
+bool Document::IsDBCSTrailByteNoExcept(char ch) const noexcept {
+	const unsigned char trail = ch;
+	switch (dbcsCodePage) {
+	case 932:
+		// Shift_jis
+		return (trail != 0x7F) &&
+			((trail >= 0x40) && (trail <= 0xFC));
+	case 936:
+		// GBK
+		return (trail != 0x7F) &&
+			((trail >= 0x40) && (trail <= 0xFE));
+	case 949:
+		// Korean Wansung KS C-5601-1987
+		return
+			((trail >= 0x41) && (trail <= 0x5A)) ||
+			((trail >= 0x61) && (trail <= 0x7A)) ||
+			((trail >= 0x81) && (trail <= 0xFE));
+	case 950:
+		// Big5
+		return
+			((trail >= 0x40) && (trail <= 0x7E)) ||
+			((trail >= 0xA1) && (trail <= 0xFE));
+	case 1361:
+		// Korean Johab KS C-5601-1992
+		return
+			((trail >= 0x31) && (trail <= 0x7E)) ||
+			((trail >= 0x81) && (trail <= 0xFE));
+	}
+	return false;
+}
+
 bool Document::IsDBCSLeadByteInvalid(char ch) const noexcept {
 	const unsigned char lead = ch;
 	switch (dbcsCodePage) {
diff --git a/src/Document.h b/src/Document.h
index 09d5841ed..88392c1a2 100644
--- a/src/Document.h
+++ b/src/Document.h
@@ -329,6 +329,7 @@ public:
 	int SCI_METHOD CodePage() const override;
 	bool SCI_METHOD IsDBCSLeadByte(char ch) const override;
 	bool IsDBCSLeadByteNoExcept(char ch) const noexcept;
+	bool IsDBCSTrailByteNoExcept(char ch) const noexcept;
 	bool IsDBCSLeadByteInvalid(char ch) const noexcept;
 	bool IsDBCSTrailByteInvalid(char ch) const noexcept;
 	int DBCSDrawBytes(std::string_view text) const noexcept;
diff --git a/test/unit/testDocument.cxx b/test/unit/testDocument.cxx
index cc6255caa..cecd14920 100644
--- a/test/unit/testDocument.cxx
+++ b/test/unit/testDocument.cxx
@@ -120,4 +120,16 @@ TEST_CASE("Document") {
 		REQUIRE(width == 1);
 		REQUIRE(ch == '=');
 	}
+
+	SECTION("CheckTrailBytes") {
+		Document doc(DocumentOption::Default);
+		const int pages[] = { 932, 936, 949, 950, 1361 };
+		for (const int page : pages) {
+			doc.SetDBCSCodePage(page);
+			for (int byteVal = 0; byteVal < 0x100; byteVal++) {
+				char ch = static_cast<char>(byteVal);
+				REQUIRE(doc.IsDBCSTrailByteNoExcept(ch) != doc.IsDBCSTrailByteInvalid(ch));
+			}
+		}
+	}
 }