add CharClassifier::GetCharsOfClass

add SCI_GETWORDCHARS(<unused>, stringresult chars) command to get word chars
add SCI_GETWHITESPACECHARS(<unused>, stringresult chars) to get whitespace chars
add SCI_GETPUNCTUATIONCHARS(<unused>, stringresult chars) to get punctutation
also add tests for {Set,Get}{Word,Whitespce,Punctuation}Chars, CharClassifier

5 files changed, 38 insertions, 0 deletions

diff --git a/src/CharClassify.cxx b/src/CharClassify.cxx
index c16af4547..7e3db737e 100644
--- a/src/CharClassify.cxx
+++ b/src/CharClassify.cxx
@@ -46,3 +46,19 @@ void CharClassify::SetCharClasses(const unsigned char *chars, cc newCharClass) {
 		}
 	}
 }
+
+int CharClassify::GetCharsOfClass(cc characterClass, unsigned char *buffer) {
+	// Get characters belonging to the given char class; return the number
+	// of characters (if the buffer is NULL, don't write to it).
+	int count = 0;
+	for (int ch = maxChar - 1; ch >= 0; --ch) {
+		if (charClass[ch] == characterClass) {
+			++count;
+			if (buffer) {
+				*buffer = static_cast<unsigned char>(ch);
+				buffer++;
+			}
+		}
+	}
+	return count;
+}
diff --git a/src/CharClassify.h b/src/CharClassify.h
index e8b798ecb..5d2734c00 100644
--- a/src/CharClassify.h
+++ b/src/CharClassify.h
@@ -19,6 +19,7 @@ public:
 	enum cc { ccSpace, ccNewLine, ccWord, ccPunctuation };
 	void SetDefaultCharClasses(bool includeWordClass);
 	void SetCharClasses(const unsigned char *chars, cc newCharClass);
+	int GetCharsOfClass(cc charClass, unsigned char *buffer);
 	cc GetClass(unsigned char ch) const { return static_cast<cc>(charClass[ch]);}
 	bool IsWord(unsigned char ch) const { return static_cast<cc>(charClass[ch]) == ccWord;}
 
diff --git a/src/Document.cxx b/src/Document.cxx
index 244e96e4f..82b9e070b 100644
--- a/src/Document.cxx
+++ b/src/Document.cxx
@@ -1615,6 +1615,10 @@ void Document::SetCharClasses(const unsigned char *chars, CharClassify::cc newCh
     charClass.SetCharClasses(chars, newCharClass);
 }
 
+int Document::GetCharsOfClass(CharClassify::cc characterClass, unsigned char *buffer) {
+    return charClass.GetCharsOfClass(characterClass, buffer);
+}
+
 void Document::SetStylingBits(int bits) {
 	stylingBits = bits;
 	stylingBitsMask = (1 << stylingBits) - 1;
diff --git a/src/Document.h b/src/Document.h
index 7e03f3d9e..30c6aee1c 100644
--- a/src/Document.h
+++ b/src/Document.h
@@ -364,6 +364,7 @@ public:
 
 	void SetDefaultCharClasses(bool includeWordClass);
 	void SetCharClasses(const unsigned char *chars, CharClassify::cc newCharClass);
+	int GetCharsOfClass(CharClassify::cc charClass, unsigned char *buffer);
 	void SetStylingBits(int bits);
 	void SCI_METHOD StartStyling(int position, char mask);
 	bool SCI_METHOD SetStyleFor(int length, char style);
diff --git a/src/Editor.cxx b/src/Editor.cxx
index d72ff302c..2bc89ba1f 100644
--- a/src/Editor.cxx
+++ b/src/Editor.cxx
@@ -7510,6 +7510,9 @@ sptr_t Editor::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lParam) {
 		caret.period = wParam;
 		break;
 
+	case SCI_GETWORDCHARS:
+		return pdoc->GetCharsOfClass(CharClassify::ccWord, reinterpret_cast<unsigned char *>(lParam));
+
 	case SCI_SETWORDCHARS: {
 			pdoc->SetDefaultCharClasses(false);
 			if (lParam == 0)
@@ -7518,6 +7521,9 @@ sptr_t Editor::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lParam) {
 		}
 		break;
 
+	case SCI_GETWHITESPACECHARS:
+		return pdoc->GetCharsOfClass(CharClassify::ccSpace, reinterpret_cast<unsigned char *>(lParam));
+
 	case SCI_SETWHITESPACECHARS: {
 			if (lParam == 0)
 				return 0;
@@ -7525,6 +7531,16 @@ sptr_t Editor::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lParam) {
 		}
 		break;
 
+	case SCI_GETPUNCTUATIONCHARS:
+		return pdoc->GetCharsOfClass(CharClassify::ccPunctuation, reinterpret_cast<unsigned char *>(lParam));
+
+	case SCI_SETPUNCTUATIONCHARS: {
+			if (lParam == 0)
+				return 0;
+			pdoc->SetCharClasses(reinterpret_cast<unsigned char *>(lParam), CharClassify::ccPunctuation);
+		}
+		break;
+
 	case SCI_SETCHARSDEFAULT:
 		pdoc->SetDefaultCharClasses(true);
 		break;