Fixed CORE-6158 - substring similar - extra characters in the result …

…for non latin characters.

src/dsql/ExprNodes.cpp

@@ -11895,18 +11895,17 @@ dsc* SubstringSimilarNode::execute(thread_db* tdbb, jrd_req* request) const
 
 	if (evaluator->result())
 	{
-		// Get the byte bounds of the matched substring.
-		unsigned start = 0;
-		unsigned length = 0;
+		// Get the character bounds of the matched substring.
+		unsigned start, length;
 		evaluator->getResultInfo(&start, &length);
 
 		dsc desc;
 		desc.makeText((USHORT) exprLen, textType);
 		EVL_make_value(tdbb, &desc, impure);
 
-		// And return it.
-		memcpy(impure->vlu_desc.dsc_address, exprStr + start, length);
-		impure->vlu_desc.dsc_length = length;
+		impure->vlu_desc.dsc_length = charSet->substring(exprLen, exprStr,
+			impure->vlu_desc.dsc_length, impure->vlu_desc.dsc_address,
+			start, length);
 
 		return &impure->vlu_desc;
 	}

src/jrd/Collation.cpp

@@ -311,7 +311,18 @@ class Re2SubstringSimilarMatcher : public BaseSubstringSimilarMatcher
 		if (textType->getAttributes() & TEXTTYPE_ATTR_ACCENT_INSENSITIVE)
 			UnicodeUtil::utf8Normalize(*bufferPtr);
 
-		return regex->matches((const char*) bufferPtr->begin(), bufferPtr->getCount(), &resultStart, &resultLength);
+		if (!regex->matches((const char*) bufferPtr->begin(), bufferPtr->getCount(), &resultStart, &resultLength))
+			return false;
+
+		if (charSetId != CS_NONE && charSetId != CS_BINARY)
+		{
+			// Get the character positions in the utf-8 string.
+			auto utf8CharSet = IntlUtil::getUtf8CharSet();
+			resultLength = utf8CharSet->length(resultLength, bufferPtr->begin() + resultStart, true);
+			resultStart = utf8CharSet->length(resultStart, bufferPtr->begin(), true);
+		}
+
+		return true;
 	}
 
 	virtual void getResultInfo(unsigned* start, unsigned* length)