New case insensitive searching implementation uses objects implementing

the CaseFolder interface to fold both search text and document text
so they can be compared with a simple strcmp.
A simple table based folder CaseFolderTable is used for 8 bit encodings
and maps input bytes to folded bytes. For multi-byte encodings
except for UTF-8 a null (output same as input) CaseFolderTable is used.
For UTF-8, more complex subclasses are used which call platform APIs
to perform the folding.
Folding is approximately to lower case although this differs between
platforms.

1 files changed, 59 insertions, 0 deletions

diff --git a/gtk/ScintillaGTK.cxx b/gtk/ScintillaGTK.cxx
index f4e762832..b53b46691 100644
--- a/gtk/ScintillaGTK.cxx
+++ b/gtk/ScintillaGTK.cxx
@@ -195,6 +195,7 @@ private:
 	void NotifyKey(int key, int modifiers);
 	void NotifyURIDropped(const char *list);
 	const char *CharacterSetID() const;
+	virtual CaseFolder *CaseFolderForEncoding();
 	virtual std::string CaseMapString(const std::string &s, int caseMapping);
 	virtual int KeyDefault(int key, int modifiers);
 	virtual void CopyToClipboard(const SelectionText &selectedText);
@@ -1338,6 +1339,64 @@ const char *ScintillaGTK::CharacterSetID() const {
 	return ::CharacterSetID(vs.styles[STYLE_DEFAULT].characterSet);
 }
 
+class CaseFolderUTF8 : public CaseFolderTable {
+public:
+	CaseFolderUTF8() {
+		StandardASCII();
+	}
+	virtual size_t Fold(char *folded, size_t sizeFolded, const char *mixed, size_t lenMixed) {
+		if ((lenMixed == 1) && (sizeFolded > 0)) {
+			folded[0] = mapping[static_cast<unsigned char>(mixed[0])];
+			return 1;
+		} else {
+			gchar *mapped = g_utf8_casefold(mixed, lenMixed);
+			size_t lenMapped = strlen(mapped);
+			if (lenMapped < sizeFolded) {
+				memcpy(folded, mapped,  lenMapped);
+			} else {
+				lenMapped = 0;
+			}
+			g_free(mapped);
+			return lenMapped;
+		}
+	}
+};
+
+CaseFolder *ScintillaGTK::CaseFolderForEncoding() {
+	if (pdoc->dbcsCodePage == SC_CP_UTF8) {
+		return new CaseFolderUTF8();
+	} else {
+		CaseFolderTable *pcf = new CaseFolderTable();
+		const char *charSetBuffer = CharacterSetID();
+		if ((pdoc->dbcsCodePage == 0) && charSetBuffer) {
+			pcf->StandardASCII();
+			// Only for single byte encodings
+			for (int i=0x80; i<0x100; i++) {
+				char sCharacter[2] = "A";
+				sCharacter[0] = i;
+				int convertedLength = 1;
+				const char *sUTF8 = ConvertText(&convertedLength, sCharacter, 1,
+					"UTF-8", charSetBuffer, false);
+				if (sUTF8) {
+					gchar *mapped = g_utf8_casefold(sUTF8, strlen(sUTF8));
+					if (mapped) {
+						int mappedLength = strlen(mapped);
+						const char *mappedBack = ConvertText(&mappedLength, mapped,
+							mappedLength, charSetBuffer, "UTF-8", false);
+						if (mappedBack && (strlen(mappedBack) == 1) && (mappedBack[0] != sCharacter[0])) {
+							pcf->SetTranslation(sCharacter[0], mappedBack[0]);
+						}
+						delete []mappedBack;
+						g_free(mapped);
+					}
+				}
+				delete []sUTF8;
+			}
+		}
+		return pcf;
+	}
+}
+
 std::string ScintillaGTK::CaseMapString(const std::string &s, int caseMapping) {
 #if GTK_MAJOR_VERSION < 2
 	return Editor::CaseMapString(s, caseMapping);