Merge pull request #44 from andreas-jonsson/more_optimize_sort

[BUG_FIXED] Fix issues related to sort optimization.

PowerEditor/src/MISC/Common/Common.cpp

@@ -718,7 +718,7 @@ generic_string stringReplace(generic_string subject, const generic_string& searc
 	return subject;
 }
 
-std::vector<generic_string> stringSplit(const generic_string& input, generic_string delimiter)
+std::vector<generic_string> stringSplit(const generic_string& input, const generic_string& delimiter)
 {
 	auto start = 0U;
 	auto end = input.find(delimiter);
@@ -734,7 +734,7 @@ std::vector<generic_string> stringSplit(const generic_string& input, generic_str
 	return output;
 }
 
-generic_string stringJoin(const std::vector<generic_string> &strings, generic_string separator)
+generic_string stringJoin(const std::vector<generic_string>& strings, const generic_string& separator)
 {
 	generic_string joined;
 	size_t length = strings.size();
@@ -748,3 +748,132 @@ generic_string stringJoin(const std::vector<generic_string> &strings, generic_st
 	}
 	return joined;
 }
+
+long long stollStrict(const generic_string& input)
+{
+	if (input.empty())
+	{
+		throw std::invalid_argument("Empty input.");
+	}
+	else
+	{
+		// Check minus characters.
+		const int minuses = std::count(input.begin(), input.end(), TEXT('-'));
+		if (minuses > 1)
+		{
+			throw std::invalid_argument("More than one minus sign.");
+		}
+		else if (minuses == 1 && input[0] != TEXT('-'))
+		{
+			throw std::invalid_argument("Minus sign must be first.");
+		}
+
+		// Check for other characters which are not allowed.
+		if (input.find_first_not_of(TEXT("-0123456789")) != std::string::npos)
+		{
+			throw std::invalid_argument("Invalid character found.");
+		}
+
+		return std::stoll(input);
+	}
+}
+
+bool allLinesAreNumericOrEmpty(const std::vector<generic_string>& lines)
+{
+	for (const generic_string& line : lines)
+	{
+		try
+		{
+			if (!line.empty())
+			{
+				stollStrict(line);
+			}
+		}
+		catch (std::invalid_argument&)
+		{
+			return false;
+		}
+		catch (std::out_of_range&)
+		{
+			return false;
+		}
+	}
+	return true;
+}
+
+std::vector<generic_string> repeatString(const generic_string& text, const size_t count)
+{
+	std::vector<generic_string> output;
+	output.reserve(count);
+	for (size_t i = 0; i < count; ++i)
+	{
+		output.push_back(text);
+	}
+	assert(output.size() == count);
+	return output;
+}
+
+std::vector<generic_string> lexicographicSort(std::vector<generic_string> input, bool isDescending)
+{
+	std::sort(input.begin(), input.end(), [isDescending](generic_string a, generic_string b)
+	{
+		if (isDescending)
+		{
+			return a.compare(b) > 0;
+		}
+		else
+		{
+			return a.compare(b) < 0;
+		}
+	});
+	return input;
+}
+
+std::vector<generic_string> numericSort(std::vector<generic_string> input, bool isDescending)
+{
+	// Pre-condition: all strings in "input" are either empty or convertible to int with stoiStrict.
+	// Note that empty lines are filtered out and added back manually to the output at the end.
+	std::vector<long long> nonEmptyInputAsNumbers;
+	size_t nofEmptyLines = 0;
+	nonEmptyInputAsNumbers.reserve(input.size());
+	for (const generic_string& line : input)
+	{
+		if (line.empty())
+		{
+			++nofEmptyLines;
+		}
+		else
+		{
+			nonEmptyInputAsNumbers.push_back(stollStrict(line));
+		}
+	}
+	assert(nonEmptyinputAsInts.size() + nofEmptyLines == input.size());
+	std::sort(nonEmptyInputAsNumbers.begin(), nonEmptyInputAsNumbers.end(), [isDescending](long long a, long long b)
+	{
+		if (isDescending)
+		{
+			return a > b;
+		}
+		else
+		{
+			return a < b;
+		}
+	});
+	std::vector<generic_string> output;
+	output.reserve(input.size());
+	const std::vector<generic_string> empties = repeatString(TEXT(""), nofEmptyLines);
+	if (!isDescending)
+	{
+		output.insert(output.end(), empties.begin(), empties.end());
+	}
+	for (const long long& sortedNumber : nonEmptyInputAsNumbers)
+	{
+		output.push_back(std::to_wstring(sortedNumber));
+	}
+	if (isDescending)
+	{
+		output.insert(output.end(), empties.begin(), empties.end());
+	}
+	assert(output.size() == input.size());
+	return output;
+}

PowerEditor/src/MISC/Common/Common.h

@@ -188,7 +188,13 @@ generic_string PathAppend(generic_string &strDest, const generic_string & str2ap
 COLORREF getCtrlBgColor(HWND hWnd);
 generic_string stringToUpper(generic_string strToConvert);
 generic_string stringReplace(generic_string subject, const generic_string& search, const generic_string& replace);
-std::vector<generic_string> stringSplit(const generic_string& input, generic_string delimiter);
+std::vector<generic_string> stringSplit(const generic_string& input, const generic_string& delimiter);
-generic_string stringJoin(const std::vector<generic_string> &strings, generic_string separator);
+generic_string stringJoin(const std::vector<generic_string>& strings, const generic_string& separator);
+long long stollStrict(const generic_string& input);
+bool allLinesAreNumericOrEmpty(const std::vector<generic_string>& lines);
+std::vector<generic_string> repeatString(const generic_string& text, const size_t count);
+
+std::vector<generic_string> numericSort(std::vector<generic_string> input, bool isDescending);
+std::vector<generic_string> lexicographicSort(std::vector<generic_string> input, bool isDescending);
 
 #endif //M30_IDE_COMMUN_H

PowerEditor/src/NppCommands.cpp

@@ -375,7 +375,7 @@ void Notepad_plus::command(int id)
 			}
 
 			_pEditView->execute(SCI_BEGINUNDOACTION);
-			_pEditView->quickSortLines(fromLine, toLine, id == IDM_EDIT_SORTLINES_DESCENDING);
+			_pEditView->sortLines(fromLine, toLine, id == IDM_EDIT_SORTLINES_DESCENDING);
 			_pEditView->execute(SCI_ENDUNDOACTION);
 
 			if (hasSelection) // there was 1 selection, so we restore it

PowerEditor/src/ScitillaComponent/ScintillaEditView.cpp

@@ -1703,7 +1703,7 @@ generic_string ScintillaEditView::getGenericTextAsString(int start, int end) con
 {
 	assert(end > start);
 	const int bufSize = end - start + 1;
-	_TCHAR *buf = new _TCHAR[bufSize];
+	TCHAR *buf = new TCHAR[bufSize];
 	getGenericText(buf, bufSize, start, end);
 	generic_string text = buf;
 	delete[] buf;
@@ -2947,7 +2947,7 @@ void ScintillaEditView::insertNewLineBelowCurrentLine()
 	execute(SCI_SETEMPTYSELECTION, execute(SCI_POSITIONFROMLINE, current_line + 1));
 }
 
-void ScintillaEditView::quickSortLines(size_t fromLine, size_t toLine, bool isDescending)
+void ScintillaEditView::sortLines(size_t fromLine, size_t toLine, bool isDescending)
 {
 	if (fromLine >= toLine)
 	{
@@ -2958,19 +2958,35 @@ void ScintillaEditView::quickSortLines(size_t fromLine, size_t toLine, bool isDe
 	const int endPos = execute(SCI_POSITIONFROMLINE, toLine) + execute(SCI_LINELENGTH, toLine);
 	const generic_string text = getGenericTextAsString(startPos, endPos);
 	std::vector<generic_string> splitText = stringSplit(text, getEOLString());
-	std::sort(splitText.begin(), splitText.end(), [isDescending](generic_string a, generic_string b)
+	const size_t lineCount = execute(SCI_GETLINECOUNT);
+	const bool sortEntireDocument = toLine == lineCount - 1;
+	if (!sortEntireDocument)
 	{
-		if (isDescending)
+		if (splitText.rbegin()->empty())
 		{
-			return a.compare(b) > 0;
+			splitText.pop_back();
+		}
+	}
+	assert(toLine - fromLine + 1 == splitText.size());
+	const bool isNumericSort = allLinesAreNumericOrEmpty(splitText);
+	std::vector<generic_string> sortedText;
+	if (isNumericSort)
+	{
+		sortedText = numericSort(splitText, isDescending);
 	}
 	else
 	{
-			return a.compare(b) < 0;
+		sortedText = lexicographicSort(splitText, isDescending);
 	}
-	});
+	const generic_string joined = stringJoin(sortedText, getEOLString());
-	const generic_string joined = stringJoin(splitText, getEOLString());
+	if (sortEntireDocument)
+	{
 		replaceTarget(joined.c_str(), startPos, endPos);
+	}
+	else
+	{
+		replaceTarget((joined + getEOLString()).c_str(), startPos, endPos);
+	}
 }
 
 bool ScintillaEditView::isTextDirectionRTL() const

PowerEditor/src/ScitillaComponent/ScintillaEditView.h

@@ -636,7 +636,7 @@ public:
 	};
 	void scrollPosToCenter(int pos);
 	generic_string getEOLString();
-	void quickSortLines(size_t fromLine, size_t toLine, bool isDescending);
+	void sortLines(size_t fromLine, size_t toLine, bool isDescending);
 	void changeTextDirection(bool isRTL);
 	bool isTextDirectionRTL() const;