mighty

commit 1db4b03a50474862ea143ad3bbe586f3feb03dd9
parent c3b65f7f8688d268cac9e824c14428d867a56286
Author: citbl <citbl@citbl.org>
Date:   Sat, 23 May 2026 23:27:55 +1000

lexer line/col

Diffstat:
M
mtcl/lexer.lua
 | 
148
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-----------
M
mtcl/main.lua
 | 
5
+++--
M
mtcl/makefile
 | 
3
+++
A
mtcl/parser.lua
 | 
3
+++

4 files changed, 137 insertions(+), 22 deletions(-)
diff --git a/mtcl/lexer.lua b/mtcl/lexer.lua
@@ -19,6 +19,10 @@ function alpha_num(c)
 	return is_alpha(c) or is_digit(c)
 end
 
+function is_space(c)
+	return c == "\t" or c == "\n" or c == " " or c == "\r"
+end
+
 function read_ident(start, src)
 	local word = ""
 	local i = start
@@ -67,20 +71,31 @@ TK = {
 	EQEQ = "equality",
 	L_PAREN = "l paren",
 	R_PAREN = "r paren",
-	DBL_QUOTE = "dbl quote",
-	SGL_QUOTE = "sgl quote",
-	LIT_FLOAT = "float lit",
-	LIT_INT = "int lit",
+	DBL_QUOTE = "dbl quo",
+	SGL_QUOTE = "sgl quo",
+	LIT_FLOAT = "a float",
+	LIT_INT = "a int",
 }
 
-Token = { kind = TK.IDK }
+Token = { kind = TK.IDK, file = "", line = 0, col = 0 }
 
 local function print_token(t)
-	print("type: " .. t.kind .. "\t value: " .. t.lexeme)
+	print(
+		t.file
+			.. " L"
+			.. t.line
+			.. ":"
+			.. t.col
+			.. " \t type: "
+			.. t.kind
+			.. "\t value: "
+			.. t.lexeme
+	)
 end
 
-function lexer:lex(src)
+function lexer:lex(file, src)
 	local i = 1
+	local line, col = 1, 1
 	local tokens = {}
 
 	local function next()
@@ -92,46 +107,139 @@ function lexer:lex(src)
 	while i <= #src do
 		local c = src:sub(i, i)
 
-		if c == "\t" or c == "\n" or c == " " or c == "\r" then
-			-- nothing
+		if is_space(c) then
+			-- skip...
+			if c == "\n" then
+				col = 1
+				line = line + 1
+			else
+				col = col + 1
+			end
 		elseif is_alpha(c:byte()) then
 			word, i = read_ident(i, src)
-			tokens[#tokens + 1] = { kind = TK.IDENT, lexeme = word }
+			col = i
+			tokens[#tokens + 1] = {
+				kind = TK.IDENT,
+				lexeme = word,
+				file = file,
+				line = line,
+				col = col,
+			}
 		elseif is_digit(c:byte()) then
 			number, i, is_float = read_number(i, src)
+			col = i
 			local kind = TK.IDK
 			if is_float then
 				kind = TK.LIT_FLOAT
 			else
 				kind = TK.LIT_INT
 			end
-			tokens[#tokens + 1] = { kind = kind, lexeme = number }
+			tokens[#tokens + 1] = {
+				kind = kind,
+				lexeme = number,
+				file = file,
+				line = line,
+				col = col,
+			}
 		elseif c == ":" then
 			if next() == ":" then
-				tokens[#tokens + 1] = { kind = TK.COLONCOLON, lexeme = "::" }
+				tokens[#tokens + 1] = {
+					kind = TK.COLONCOLON,
+					lexeme = "::",
+					file = file,
+					line = line,
+					col = col,
+				}
 				i = i + 1
+				col = col + 1
 			else
-				tokens[#tokens + 1] = { kind = TK.COLON, lexeme = ":" }
+				tokens[#tokens + 1] = {
+					kind = TK.COLON,
+					lexeme = ":",
+					file = file,
+					line = line,
+					col = col,
+				}
 			end
 		elseif c == "=" then
 			if next() == "=" then
-				tokens[#tokens + 1] = { kind = TK.EQEQ, lexeme = "==" }
+				tokens[#tokens + 1] = {
+					kind = TK.EQEQ,
+					lexeme = "==",
+
+					file = file,
+					line = line,
+					col = col,
+				}
 				i = i + 1
+				col = col + 1
 			else
-				tokens[#tokens + 1] = { kind = TK.EQ, lexeme = "=" }
+				tokens[#tokens + 1] = {
+					kind = TK.EQ,
+					lexeme = "=",
+
+					file = file,
+					line = line,
+					col = col,
+				}
 			end
 		elseif c == "(" then
-			tokens[#tokens + 1] = { kind = TK.L_PAREN, lexeme = "(" }
+			tokens[#tokens + 1] = {
+				kind = TK.L_PAREN,
+				lexeme = "(",
+
+				file = file,
+				line = line,
+				col = col,
+			}
+			i = i + 1
+			col = col + 1
 		elseif c == ")" then
-			tokens[#tokens + 1] = { kind = TK.R_PAREN, lexeme = ")" }
+			tokens[#tokens + 1] = {
+				kind = TK.R_PAREN,
+				lexeme = ")",
+
+				file = file,
+				line = line,
+				col = col,
+			}
+			i = i + 1
+			col = col + 1
 		elseif c == "'" then
-			tokens[#tokens + 1] = { kind = TK.SGL_QUOTE, lexeme = "'" }
+			tokens[#tokens + 1] = {
+				kind = TK.SGL_QUOTE,
+				lexeme = "'",
+
+				file = file,
+				line = line,
+				col = col,
+			}
+			i = i + 1
+			col = col + 1
 		elseif c == '"' then
-			tokens[#tokens + 1] = { kind = TK.DBL_QUOTE, lexeme = '"' }
+			tokens[#tokens + 1] = {
+				kind = TK.DBL_QUOTE,
+				lexeme = '"',
+
+				file = file,
+				line = line,
+				col = col,
+			}
+			i = i + 1
+			col = col + 1
 		else
-			tokens[#tokens + 1] = { kind = TK.IDK, lexeme = "" }
+			tokens[#tokens + 1] = {
+				kind = TK.IDK,
+				lexeme = "",
+				file = file,
+				line = line,
+				col = col,
+			}
+			i = i + 1
+			col = col + 1
 		end
 		i = i + 1
+		col = col + 1
 	end
 
 	for i = 1, #tokens do
diff --git a/mtcl/main.lua b/mtcl/main.lua
@@ -2,9 +2,10 @@
 
 local lexer = require("lexer")
 
-local file = io.open(arg[1], "rb")
+local filename = arg[1] -- TODO check
+local file = io.open(filename, "rb")
 local contents = file:read("*all")
 
-lexer:lex(contents)
+lexer:lex(filename, contents)
 
 print(#lexer.tokens .. " tokens")
diff --git a/mtcl/makefile b/mtcl/makefile
@@ -3,3 +3,6 @@ MAKEFLAGS += --silent
 default:
 	lua main.lua target.mty
 	stylua --column-width 80 .
+
+fmt:
+	stylua --column-width 80 .
diff --git a/mtcl/parser.lua b/mtcl/parser.lua
@@ -0,0 +1,3 @@
+local parser = {}
+
+function parser:parse(filename, src, parser) end