-- SPDX-FileCopyrightText: © 2026 Vladimir Zorin <vladimir@deviant.guru>
-- SPDX-License-Identifier: LicenseRef-OWL-1.0-or-later
-- Licensed under OWL v1.0+. See LICENSE.

-- C syntax highlighter for LEM.

local style = require("lem.highlight.style")

local STATE_NORMAL = nil
local STATE_BLOCK_COMMENT = "block_comment"
local STATE_PREPROCESSOR = "preprocessor"

local PUNCTUATION = style.COMMON_PUNCTUATION
local scan_quoted = style.scan_quoted
local scan_ident_end = style.scan_ident_end
local scan_number_end = style.scan_number_end

local OPERATORS = {
	["+"] = true,
	["-"] = true,
	["*"] = true,
	["/"] = true,
	["%"] = true,
	["="] = true,
	["<"] = true,
	[">"] = true,
	["!"] = true,
	["&"] = true,
	["|"] = true,
	["^"] = true,
	["~"] = true,
	["?"] = true,
	[":"] = true,
}

local LITERALS = {
	["true"] = true,
	["false"] = true,
	["NULL"] = true,
}

local highlight_line = function(self, text, state)
	local spans = {}
	local i = 1
	local len = #text

	-- Continuation of a preprocessor directive from previous line.
	-- A blank line terminates the directive.
	if state == STATE_PREPROCESSOR then
		if len == 0 then
			return spans, STATE_NORMAL
		end
		spans[#spans + 1] = { 1, len, "directive" }
		if text:match("\\%s*$") then
			return spans, STATE_PREPROCESSOR
		end
		return spans, STATE_NORMAL
	end

	-- Inside a block comment from previous line
	if state == STATE_BLOCK_COMMENT then
		local close = text:find("%*/", i)
		if close then
			spans[#spans + 1] = { i, close + 1, "disabled" }
			i = close + 2
			state = STATE_NORMAL
		else
			spans[#spans + 1] = { i, len, "disabled" }
			return spans, STATE_BLOCK_COMMENT
		end
	end

	-- Preprocessor directive; trailing backslash extends it to the next line.
	local directive_start = text:match("^(%s*#)")
	if directive_start and i == 1 then
		spans[#spans + 1] = { 1, len, "directive" }
		if text:match("\\%s*$") then
			return spans, STATE_PREPROCESSOR
		end
		return spans, STATE_NORMAL
	end

	while i <= len do
		local ch = text:sub(i, i)

		-- Line comment: //
		if ch == "/" and text:sub(i + 1, i + 1) == "/" then
			spans[#spans + 1] = { i, len, "comment" }
			return spans, STATE_NORMAL

		-- Block comment: /*
		elseif ch == "/" and text:sub(i + 1, i + 1) == "*" then
			local close = text:find("%*/", i + 2)
			if close then
				spans[#spans + 1] = { i, close + 1, "disabled" }
				i = close + 2
			else
				spans[#spans + 1] = { i, len, "disabled" }
				return spans, STATE_BLOCK_COMMENT
			end

		-- String or char literal (double or single quote)
		elseif ch == '"' or ch == "'" then
			local j = scan_quoted(text, i, len, ch)
			spans[#spans + 1] = { i, j, "literal" }
			i = j + 1

		-- Number
		elseif ch:match("%d") or (ch == "." and i + 1 <= len and text:sub(i + 1, i + 1):match("%d")) then
			local j = scan_number_end(text, i, len, "[%d%.eEfFlLuU]")
			spans[#spans + 1] = { i, j, "literal" }
			i = j + 1

		-- Punctuation
		elseif PUNCTUATION[ch] then
			spans[#spans + 1] = { i, i, "punctuation" }
			i = i + 1

		-- Operators
		elseif OPERATORS[ch] then
			local j = i
			local next_ch = text:sub(i + 1, i + 1)
			if (ch == "=" or ch == "!" or ch == "<" or ch == ">") and next_ch == "=" then
				j = i + 1
			elseif (ch == "&" and next_ch == "&") or (ch == "|" and next_ch == "|") then
				j = i + 1
			elseif (ch == "+" and next_ch == "+") or (ch == "-" and next_ch == "-") then
				j = i + 1
			elseif ch == "-" and next_ch == ">" then
				j = i + 1
			elseif (ch == "<" and next_ch == "<") or (ch == ">" and next_ch == ">") then
				j = i + 1
			end
			spans[#spans + 1] = { i, j, "operator" }
			i = j + 1

		-- Identifiers
		elseif ch:match("[%a_]") then
			local word_end = scan_ident_end(text, i, len)
			local word = text:sub(i, word_end)

			if LITERALS[word] then
				spans[#spans + 1] = { i, word_end, "literal" }
			else
				-- Definition: name followed by ( at start of line (function definition)
				local before = text:sub(1, i - 1)
				local after = text:sub(word_end + 1)
				if before:match("^[%w_%s%*]+$") and after:match("^%s*%(") and not before:match("^%s*$") then
					-- Likely a function definition: "type name("
					-- But only if this isn't an if/while/for/return
					if
						not word:match("^if$")
						and not word:match("^while$")
						and not word:match("^for$")
						and not word:match("^return$")
						and not word:match("^switch$")
						and not word:match("^sizeof$")
					then
						spans[#spans + 1] = { i, word_end, "definition" }
					end
				end
			end
			i = word_end + 1
		else
			i = i + 1
		end
	end

	return spans, STATE_NORMAL
end

local new = function()
	return {
		highlight_line = highlight_line,
	}
end

return { new = new }