From d94cb62e0ba795b24776ac6a5e8b85bc06fac696 Mon Sep 17 00:00:00 2001
From: Lazar Bulovan <lazar.bulovan@l-tech.biz>
Date: Thu, 19 Feb 2026 12:01:33 +0100
Subject: [PATCH] fix(lsp): add more robust validation in semantic tokens

- Wrap token generation in pcall to catch any errors
- Add validation for all token parameters (line, startChar, length)
- Validate deltaStart is non-negative
- Return empty tokens if any error occurs
- This should prevent the 'Invalid line: out of range' errors
---
 src/main.lua | 147 +++++++++++++++++++++++++--------------------------
 1 file changed, 73 insertions(+), 74 deletions(-)

diff --git a/src/main.lua b/src/main.lua
index f689aa9..6abdde9 100644
--- a/src/main.lua
+++ b/src/main.lua
@@ -563,92 +563,91 @@ function handlers.textDocument_semanticTokensFull(params)
 		return { data = {} }
 	end
 
+	-- Protected token collection
 	local tokens = {}
-	local lines = {}
-	for line in doc.content:gmatch("([^\n]*)\n") do
-		table.insert(lines, line)
-	end
-	
-	local total_lines = #lines
-
-	local function add_token(line, startChar, length, tokenType)
-		-- Validate line number is within bounds
-		if line < 0 or line >= total_lines then
-			return
+	local ok, err = pcall(function()
+		local lines = {}
+		for line in doc.content:gmatch("([^\n]*)\n") do
+			table.insert(lines, line)
 		end
-		local deltaLine = line
-		local deltaStart = startChar
-		if #tokens >= 4 then
-			local prevLine = tokens[#tokens - 3]
-			local prevStart = tokens[#tokens - 2]
-			if line == prevLine then
-				deltaLine = 0
-				deltaStart = startChar - prevStart
-			else
-				deltaLine = line - prevLine
+		
+		local total_lines = #lines
+
+		local function add_token(line, startChar, length)
+			-- Validate all parameters
+			if not line or not startChar or not length then
+				return
 			end
+			if line < 0 or line >= total_lines then
+				return
+			end
+			if startChar < 0 or length <= 0 then
+				return
+			end
+			if startChar + length > 10000 then
+				return
+			end
+			
+			local deltaLine = line
+			local deltaStart = startChar
+			if #tokens >= 4 then
+				local prevLine = tokens[#tokens - 3]
+				local prevStart = tokens[#tokens - 2]
+				if line == prevLine then
+					deltaLine = 0
+					deltaStart = startChar - prevStart
+				else
+					deltaLine = line - prevLine
+				end
+				-- Validate deltaStart is non-negative
+				if deltaStart < 0 then
+					return
+				end
+			end
+			table.insert(tokens, deltaLine)
+			table.insert(tokens, deltaStart)
+			table.insert(tokens, length)
+			table.insert(tokens, 0)
+			table.insert(tokens, 0)
 		end
-		table.insert(tokens, deltaLine)
-		table.insert(tokens, deltaStart)
-		table.insert(tokens, length)
-		table.insert(tokens, 0)
-		table.insert(tokens, 0)
-	end
 
-	for i, line in ipairs(lines) do
-		local lineNum = i - 1
-		local trimmed = line:gsub("^%s+", "")
+		for i, line in ipairs(lines) do
+			local lineNum = i - 1
+			local trimmed = line:gsub("^%s+", "")
 
-		if doc.variables then
-			for var_name, info in pairs(doc.variables) do
-				-- Validate variable line number is within bounds
-				if info.line and info.line >= 0 and info.line < total_lines and info.line == lineNum then
-					local startChar = line:find(var_name, 1, true)
-					if startChar then
-						add_token(lineNum, startChar - 1, #var_name, "variable")
+			if doc.variables then
+				for var_name, info in pairs(doc.variables) do
+					if info and info.line and info.line >= 0 and info.line < total_lines and info.line == lineNum then
+						local startChar = line:find(var_name, 1, true)
+						if startChar then
+							add_token(lineNum, startChar - 1, #var_name)
+						end
 					end
 				end
 			end
-		end
 
-		local keywords = {
-			"IF",
-			"THEN",
-			"ELSIF",
-			"ELSE",
-			"END_IF",
-			"FOR",
-			"TO",
-			"BY",
-			"DO",
-			"END_FOR",
-			"WHILE",
-			"END_WHILE",
-			"REPEAT",
-			"UNTIL",
-			"END_REPEAT",
-			"CASE",
-			"OF",
-			"END_CASE",
-			"VAR_INPUT",
-			"VAR_OUTPUT",
-			"VAR_IN_OUT",
-			"VAR_TEMP",
-			"VAR",
-			"END_VAR",
-			"BEGIN",
-			"EXIT",
-			"CONTINUE",
-			"RETURN",
-			"TRUE",
-			"FALSE",
-		}
-		for _, kw in ipairs(keywords) do
-			local startPos = line:find("%f[%a]" .. kw .. "%f[%A]")
-			if startPos then
-				add_token(lineNum, startPos - 1, #kw, "keyword")
+			local keywords = {
+				"IF", "THEN", "ELSIF", "ELSE", "END_IF",
+				"FOR", "TO", "BY", "DO", "END_FOR",
+				"WHILE", "END_WHILE",
+				"REPEAT", "UNTIL", "END_REPEAT",
+				"CASE", "OF", "END_CASE",
+				"VAR_INPUT", "VAR_OUTPUT", "VAR_IN_OUT", "VAR_TEMP", "VAR", "END_VAR",
+				"BEGIN", "EXIT", "CONTINUE", "RETURN",
+				"TRUE", "FALSE",
+			}
+			for _, kw in ipairs(keywords) do
+				local startPos = line:find("%f[%a]" .. kw .. "%f[%A]")
+				if startPos then
+					add_token(lineNum, startPos - 1, #kw)
+				end
 			end
 		end
+	end)
+	
+	-- If any error occurred, return empty tokens
+	if not ok then
+		return { data = {} }
 	end
 
 	return { data = tokens }