[BUG] Smart indentation (Advanced mode) occasionally adds an extra TAB character of indentation under certain conditions

[softmgr]

Is there an existing issue for this?

• I have searched the existing issues

Is the issue reproducible in Notepad++ without plugin?

• I have tried Notepad++ without plugin

Is the issue reproducible in portable version of Notepad++?

• I have tried portable version of Notepad++

Does the issue affect SciTE?

• I have tried SciTE

Description of the Issue

I encountered a situation (there might be others): In a .cpp file, when an if or while statement does not have curly braces, and you type the { character on the next line, it incorrectly adds an extra TAB character of indentation.

@donho The Smart Indentation (Advanced Indentation) feature has many bugs based on my testing. Many situations are not considered, such as brace matching errors, indentation errors when braces are missing, and failing to ignore comments in the code, etc.
Here is a sample file: sample.cpp.zip

Steps To Reproduce

1. Paste the following code into the editor and set the language to "C++".

	void function(const wchar_t* str)
	{
		if (str)
			return str;


	}

2. On line 5, type the { character. This will automatically add 3 levels of indentation, and the file content will look like this:

	void function(const wchar_t* str)
	{
		if (str)
			return str;
			{

	}

3. Paste the following code into the editor and set the language to "C++".

	void function(const wchar_t* str)
	{
		if (str) {



	}

4. Press Enter to the right of if (str). This will automatically add 3 levels of indentation, and the file content will look like this:

	void function(const wchar_t* str)
	{
		if (str)
			{



	}

5. Paste the following code into the editor and set the language to "C++".

	void function(const wchar_t* str)
	{
		if (str) { /*This is a comment*/



	}

6. Press Enter to the right of /*This is a comment*/. This will automatically add 2 levels of indentation, and the file content will look like this:

	void function(const wchar_t* str)
	{
		if (str) { /*This is a comment*/
		



	}

Current Behavior

In these specific cases, there will be too many or too few TAB characters.

Expected Behavior

The indentation should follow the standards used by software like Visual Studio and others.

Debug Information

Notepad++ v8.9.2   (64-bit)
Build time: Feb 13 2026 - 02:47:53
Scintilla/Lexilla included: 5.5.8/5.4.6
Boost Regex included: 1_90
pugixml included: 1.15
nlohmann JSON included: 3.12.0
Path: C:\Program Files\Notepad++\notepad++.exe
Command Line: 
Admin mode: ON
Local Conf mode: OFF
Cloud Config: OFF
Periodic Backup: ON
Placeholders: OFF
Scintilla Rendering Mode: SC_TECHNOLOGY_DIRECTWRITE (1)
Multi-instance Mode: monoInst
asNotepad: OFF
File Status Auto-Detection: cdEnabledNew (for current file/tab only)
Dark Mode: OFF
Display Info:
    primary monitor: 1920x1200, scaling 100%
    visible monitors count: 1
    installed Display Class adapters: 
        0000: Description - Intel(R) Graphics
        0000: DriverVersion - 32.0.101.8247
OS Name: Windows 11 Enterprise (64-bit)
OS Version: 23H2
OS Build: 22631.6783
Current ANSI codepage: 1252
Plugins: 
    mimeTools (3.1)

Anything else?

No response

[softmgr]

@donho
I have completed part of the modifications, but there is still a long way to go before achieving the final perfect result. The completed additions and fixes in the following code are:

1. Added the isCLike function to improve efficiency.
2. Detects whether the '{' and '}' characters are inside comments or strings.
3. Checks if there is a left curly brace '{' in the specified range.
4. Updated the findMatchedBracePos function to handle checks inside comments and strings.
5. Added automatic indentation for the ';' character (similar to the behavior in Visual Studio).

The following code is for reference only.

// Buffer.h

	int isCLike() {  // added
		switch (_lang) {
			case L_PERL:
			case L_RUST:
			case L_POWERSHELL:
			case L_JSON:
			case L_JSON5:
				return 1;

			//case L_TEXT:
			case L_C:
			case L_CPP:
			case L_JAVA:
			case L_CS:
			case L_OBJC:
			case L_PHP:
			case L_JS_EMBEDDED:
			case L_JAVASCRIPT:
			case L_JSP:
			case L_CSS:
			case L_TYPESCRIPT:
			case L_GOLANG:
			case L_SWIFT:
			case L_USER:  // added
				return 2;

			default:
				return 0;
		}
	}

// Notepad_plus.cpp

// Check if the given "position" is within a comment
bool isPositionInCommentOrQuotes(ScintillaEditView* pEditView, intptr_t rangeStart, intptr_t rangeEnd, intptr_t position)
{
	//if (!pEditView || rangeStart > rangeEnd || pos < rangeStart || pos > rangeEnd)
	//	return false;

	bool isMultiLineComment = false;
	bool inQuotes = false;

	for (intptr_t i = rangeStart; i < rangeEnd; ++i)
	{
		char ch = static_cast<char>(pEditView->execute(SCI_GETCHARAT, i));

		if (!isMultiLineComment)
		{
			if (ch == '"' && (char)pEditView->execute(SCI_GETCHARAT, i - 1) != '\\')  // Check if it's inside double quotes
				inQuotes = !inQuotes;  // Toggle the double quote flag
		}

		if (!inQuotes)
		{
			// "//" and "/*"
			if (!isMultiLineComment && ch == '/')
			{
				char next = static_cast<char>(pEditView->execute(SCI_GETCHARAT, i + 1));
				if (next == '/')
				{
					if (i + 1 < position)  // position is after a single-line comment
						return true;
				}
				else if (next == '*')
				{
					isMultiLineComment = true;
					++i;
					continue;
				}
			}

			// "*/"
			if (isMultiLineComment && ch == '*')
			{
				char next = static_cast<char>(pEditView->execute(SCI_GETCHARAT, i + 1));
				if (next == '/')
				{
					isMultiLineComment = false;
					++i;
					continue;
				}
			}

			// Inside a multi-line comment
			if (isMultiLineComment && i == position)
			{
				return true;
			}
		}
		else if (i == position)  // Input of '{' or '}' inside double quotes
		{
			return true;
		}
	}

	return false;
}

// Check if there's a left curly brace '{' in the specified range
bool isLeftCurlyBraceInRange(ScintillaEditView* pEditView, intptr_t rangeStart, intptr_t rangeEnd)
{
	int count = 0;
	bool isMultiLineComment = false;
	bool inQuotes = false;
	for (intptr_t pos = rangeEnd; pos >= rangeStart; --pos)
	{
		char currentChar = (char)pEditView->execute(SCI_GETCHARAT, pos);

		if (!isMultiLineComment)
		{
			if (currentChar == '"' && (char)pEditView->execute(SCI_GETCHARAT, pos - 1) != '\\')  // Check if it's inside double quotes
				inQuotes = !inQuotes;  // Toggle the double quote flag

			if (inQuotes)  // Skip characters inside double quotes
				continue;
		}

		if (!inQuotes)
		{
			if (isMultiLineComment)  // Skip characters if currently inside a multi-line comment
			{
				if (currentChar == '*' && (char)pEditView->execute(SCI_GETCHARAT, pos - 1) == '/')
				{
					isMultiLineComment = false;
					pos--;  // Skip the start of multi-line comment characters '/'
				}
				continue;
			}

			if (currentChar == '/')  // It could be the end of a multi-line comment
			{
				char prevChr = (char)pEditView->execute(SCI_GETCHARAT, pos - 1);
				if (prevChr == '/')
				{
					pos--;  // Skip a single-line comment
					count = 0;  // Reset to 0 after processing all previous comments
					continue;
				}
				else if (prevChr == '*')
				{
					isMultiLineComment = true;
					continue;
				}
			}
		}

		if (currentChar == '{')
			++count;
		else if (currentChar == '}')
			--count;
	}

	return (count == 1);  // If there's only one '{' on the left side within the range, return true
}

intptr_t Notepad_plus::findMachedBracePos(size_t startPos, size_t endPos, char targetSymbol, char matchedSymbol)
{
	if (startPos == endPos)
		return -1;

	if (startPos > endPos) // backward
	{
		bool isMultiLineComment = false;  // added
		bool inQuotes = false;  // added
		int balance = 0;
		for (intptr_t i = startPos; i >= static_cast<intptr_t>(endPos); --i)
		{
			char aChar = static_cast<char>(_pEditView->execute(SCI_GETCHARAT, i));
			
			////////////////
			/* added */
			if (!isMultiLineComment)
			{
				if (aChar == '"' && static_cast<char>(_pEditView->execute(SCI_GETCHARAT, i - 1)) != '\\')
					inQuotes = !inQuotes;

				if (inQuotes)
					continue;
			}
			
			/* added */
			if (!inQuotes)
			{
				if (isMultiLineComment)
				{
					if (aChar == '*' && (char)_pEditView->execute(SCI_GETCHARAT, i - 1) == '/')
					{
						isMultiLineComment = false;
						i--;
					}
					continue;
				}

				if (aChar == '/')
				{
					char prevChr = (char)_pEditView->execute(SCI_GETCHARAT, i - 1);
					if (prevChr == '/')
					{
						i--;
						balance = 0;  // Reset to 0
						continue;
					}
					else if (prevChr == '*')
					{
						isMultiLineComment = true;
						continue;
					}
				}
			}
			////////////////
		

			if (aChar == targetSymbol)
			{
				if (balance == 0)
					return i;
				--balance;
			}
			else if (aChar == matchedSymbol)
			{
				++balance;
			}
		}
	}
	else // forward
	{
	}
	return -1;
}

void Notepad_plus::maintainIndentation(wchar_t ch)
{
	const NppGUI& nppGui = NppParameters::getInstance().getNppGUI();
	if (nppGui._maintainIndent == autoIndent_none)
		return;

	intptr_t eolMode = _pEditView->execute(SCI_GETEOLMODE);
	intptr_t curLine = _pEditView->getCurrentLineNumber();
	intptr_t prevLine = curLine - 1;
	intptr_t indentAmountPrevLine = 0;
	intptr_t tabWidth = _pEditView->execute(SCI_GETTABWIDTH);

	const bool isEnter = ((eolMode == SC_EOL_CRLF || eolMode == SC_EOL_LF) && ch == '\n') || (eolMode == SC_EOL_CR && ch == '\r');  // added
	// Do not alter indentation if we were at the beginning of the line and we pressed Enter
	if (isEnter && prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)  // modified
		return;

	Buffer* buf = _pEditView->getCurrentBuffer();  // added
	LangType type = buf->getLangType();  // modified  //LangType type = _pEditView->getCurrentBuffer()->getLangType();
	ExternalLexerAutoIndentMode autoIndentMode = ExternalLexerAutoIndentMode::Standard;

	// For external languages, query for custom auto-indentation funcionality
	if (type >= L_EXTERNAL)
	{
		NppParameters& nppParam = NppParameters::getInstance();
		autoIndentMode = nppParam.getELCFromIndex(type - L_EXTERNAL)->_autoIndentMode;
		if (autoIndentMode == ExternalLexerAutoIndentMode::Custom)
			return;
	}

	if (nppGui._maintainIndent == autoIndent_basic) // Basic indentation mode only
	{
		if (isEnter)  // modified
		{
			// Search the non-empty previous line
			while (prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)
				prevLine--;

			if (prevLine >= 0)
			{
				indentAmountPrevLine = _pEditView->getLineIndent(prevLine);
			}

			if (indentAmountPrevLine > 0)
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine);
			}
		}

		return;
	}

	// else nppGui._maintainIndent == autoIndent_advance

	if (buf->isCLike() >= 1 || autoIndentMode == ExternalLexerAutoIndentMode::C_Like)  // modified
	{
		if (isEnter)  // modified
		{
			// Search the non-empty previous line
			while (prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)
				prevLine--;

			// Get previous line's Indent
			if (prevLine >= 0)
			{
				indentAmountPrevLine = _pEditView->getLineIndent(prevLine);
			}

			// get previous char from current line
			intptr_t prevPos = _pEditView->execute(SCI_GETCURRENTPOS) - (eolMode == SC_EOL_CRLF ? 3 : 2);
			UCHAR prevChar = (UCHAR)_pEditView->execute(SCI_GETCHARAT, prevPos);
			auto curPos = _pEditView->execute(SCI_GETCURRENTPOS);
			UCHAR nextChar = (UCHAR)_pEditView->execute(SCI_GETCHARAT, curPos);

			if (prevChar == '{')
			{
				if (nextChar == '}')
				{
					const char *eolChars;
					if (eolMode == SC_EOL_CRLF)
						eolChars = "\r\n";
					else if (eolMode == SC_EOL_LF)
						eolChars = "\n";
					else
						eolChars = "\r";

					_pEditView->execute(SCI_INSERTTEXT, _pEditView->execute(SCI_GETCURRENTPOS), reinterpret_cast<LPARAM>(eolChars));
					_pEditView->setLineIndent(curLine + 1, indentAmountPrevLine);
				}
				_pEditView->setLineIndent(curLine, indentAmountPrevLine + tabWidth);
			}
			else if (nextChar == '{')
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine);
			}
			// These languages do no support single line control structures without braces.
			else if (buf->isCLike() == 1)  // modified
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine);
			}
			else if (isConditionExprLine(prevLine))
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine + tabWidth);
			}
			else
			{
				if (indentAmountPrevLine > 0)
				{
					if (prevLine > 0 && isConditionExprLine(prevLine - 1))
						_pEditView->setLineIndent(curLine, indentAmountPrevLine - tabWidth);
					else
						_pEditView->setLineIndent(curLine, indentAmountPrevLine);
				}
			}
		}
		else if (ch == '{')
		{
			// if no character in front of {, aligned with prev line's indentation
			auto startPos = _pEditView->execute(SCI_POSITIONFROMLINE, curLine);
			LRESULT endPos = _pEditView->execute(SCI_GETCURRENTPOS);

			if (isPositionInCommentOrQuotes(_pEditView, startPos, _pEditView->execute(SCI_GETLINEENDPOSITION, curLine), endPos - 1))  // added
				return;  // added

			for (LRESULT i = endPos - 2; i > 0 && i >= startPos; --i)
			{
				UCHAR aChar = (UCHAR)_pEditView->execute(SCI_GETCHARAT, i);
				if (aChar != ' ' && aChar != '\t')
					return;
			}

			// Search the non-empty previous line
			while (prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)
				prevLine--;

			// Get previous line's Indent
			if (prevLine >= 0)
			{
				indentAmountPrevLine = _pEditView->getLineIndent(prevLine);

				auto startPos2 = _pEditView->execute(SCI_POSITIONFROMLINE, prevLine);
				auto endPos2 = _pEditView->execute(SCI_GETLINEENDPOSITION, prevLine);
				_pEditView->execute(SCI_SETSEARCHFLAGS, SCFIND_REGEXP | SCFIND_POSIX);
				_pEditView->execute(SCI_SETTARGETRANGE, startPos2, endPos2);

				const char braceExpr[] = "[ \t]*\\{.*";

				intptr_t posFound = _pEditView->execute(SCI_SEARCHINTARGET, strlen(braceExpr), reinterpret_cast<LPARAM>(braceExpr));
				if (posFound >= 0)
				{
					auto end = _pEditView->execute(SCI_GETTARGETEND);
					if (end == endPos2)
						indentAmountPrevLine += tabWidth;
				}
			}

			_pEditView->setLineIndent(curLine, indentAmountPrevLine);

		}
		else if (ch == '}')
		{
			// Look backward for the pair {
			intptr_t startPos = _pEditView->execute(SCI_GETCURRENTPOS);

			if (isPositionInCommentOrQuotes(_pEditView, _pEditView->execute(SCI_POSITIONFROMLINE, curLine), _pEditView->execute(SCI_GETLINEENDPOSITION, curLine), startPos - 1))  // added
				return;  // added

			if (static_cast<char>(_pEditView->execute(SCI_GETCHARAT, startPos - 2)) == '\'')  // added
				return;  // added

			if (startPos != 0)
				startPos -= 1;
			intptr_t posFound = findMachedBracePos(startPos - 1, 0, '{', '}');

			// if no { found, do nothing
			if (posFound == -1)
				return;

			// if { is in the same line, do nothing
			intptr_t matchedPairLine = _pEditView->execute(SCI_LINEFROMPOSITION, posFound);
			if (matchedPairLine == curLine)
				return;

			// { is in another line, get its indentation
			indentAmountPrevLine = _pEditView->getLineIndent(matchedPairLine);

			// aligned } indent with {
			_pEditView->setLineIndent(curLine, indentAmountPrevLine);

			/*
			// indent lines from { to }
			for (int i = matchedPairLine + 1; i < curLine; ++i)
				_pEditView->setLineIndent(i, indentAmountPrevLine + tabWidth);
			*/
		}
		else if (ch == ';')  // added
		{
			intptr_t currPos = _pEditView->execute(SCI_GETCURRENTPOS);
			intptr_t lineStart = _pEditView->execute(SCI_POSITIONFROMLINE, curLine);
			intptr_t lineEnd = _pEditView->execute(SCI_GETLINEENDPOSITION, curLine);
			char leftCh = static_cast<char>(_pEditView->execute(SCI_GETCHARAT, currPos - 2));  // A character on the left of ';'
			if (isPositionInCommentOrQuotes(_pEditView, lineStart, lineEnd, currPos - 1) || leftCh == '/' || leftCh == '\\' || leftCh == '\'' || leftCh == '\r' || leftCh == '\n')
				return;

			_pEditView->execute(SCI_BEGINUNDOACTION);

			// First, remove the extra spaces and TABs to the left of the ';'
			intptr_t removeStart = -1, removeEnd = -1;
			for (intptr_t i = currPos - 2; i >= lineStart; --i)
			{
				char aChar = static_cast<char>(_pEditView->execute(SCI_GETCHARAT, i));
				bool isBlank = (aChar == ' ' || aChar == '\t');
				if (isBlank && removeEnd == -1)
				{
					removeEnd = i + 1;
				}
				else if (!isBlank && removeEnd != -1)
				{
					removeStart = i + 1;
					break;
				}
			}
			if (removeStart != -1 && removeEnd != -1)
			{
				_pEditView->execute(SCI_SETTARGETRANGE, removeStart, removeEnd);
				_pEditView->execute(SCI_REPLACETARGET, 0, reinterpret_cast<LPARAM>(""));
			}

			// Insert indentation
			while (prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)
				prevLine--;

			if (prevLine >= 0)
			{
				indentAmountPrevLine = _pEditView->getLineIndent(prevLine);
				if (isLeftCurlyBraceInRange(_pEditView, _pEditView->execute(SCI_POSITIONFROMLINE, prevLine), _pEditView->execute(SCI_GETLINEENDPOSITION, prevLine)))  // If there is a '{' character in the previous line, the current line needs to be indented
					indentAmountPrevLine += tabWidth;
				else if (!isLeftCurlyBraceInRange(_pEditView, lineStart, lineEnd))  // If there is no '{' character in the previous line and no '{' character in the current line, then the current line needs to be indented
					indentAmountPrevLine += tabWidth;
				_pEditView->setLineIndent(curLine, indentAmountPrevLine);
			}

			_pEditView->execute(SCI_ENDUNDOACTION);
		}
	}
	else if (type == L_PYTHON)
	{
		if (isEnter)  // modified
		{
			// Search the non-empty previous line
			while (prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)
				prevLine--;

			// Get previous line's Indent
			if (prevLine >= 0)
			{
				indentAmountPrevLine = _pEditView->getLineIndent(prevLine);
			}

			_pEditView->execute(SCI_SETSEARCHFLAGS, SCFIND_REGEXP | SCFIND_POSIX);

			auto startPos = _pEditView->execute(SCI_POSITIONFROMLINE, prevLine);
			auto endPos = _pEditView->execute(SCI_GETLINEENDPOSITION, prevLine);
			_pEditView->execute(SCI_SETTARGETRANGE, startPos, endPos);

			// colon optionally followed by only whitespace and/or start-of-comment, but NOT on a line that is already a comment
			const char colonExpr[] = ":[ \t]*(#|$)";

			auto posColon = _pEditView->execute(SCI_SEARCHINTARGET, strlen(colonExpr), reinterpret_cast<LPARAM>(colonExpr));

			// when colon found, additionally check that it is not in a comment, inside a string, etc.
			if ((posColon >= 0) && (_pEditView->execute(SCI_GETSTYLEINDEXAT, posColon) == SCE_P_OPERATOR))
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine + tabWidth);
			}
			else if (indentAmountPrevLine > 0)
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine);
			}
		}
	}
	else // Basic indentation mode
	{
		if (isEnter)  // modified
		{
			// Search the non-empty previous line
			while (prevLine >= 0 && _pEditView->getLineLength(prevLine) == 0)
				prevLine--;

			if (prevLine >= 0)
			{
				indentAmountPrevLine = _pEditView->getLineIndent(prevLine);
			}

			if (indentAmountPrevLine > 0)
			{
				_pEditView->setLineIndent(curLine, indentAmountPrevLine);
			}
		}
	}
}

[donho]

@softmgr
Your code is adapted in PR #17953.
However, I didn't notice any improvment.